Return to running protocols: criteria, progression and what to document

A return to running protocol is three things: the entry criteria you check before the first run, the walk-run progression you build after it, and the note that records why you let this runner go. The tissue decides the criteria. The progression barely changes from one tissue to the next, and the part most of us do worst is writing down what we actually checked.

This page is for clinicians. The patient-facing versions sit at the return to running plan and ACL return to running, and those are the pages to hand over, not this one.

What has to be true before a first run, by tissue

Four things are common to every protocol worth using. No pain at rest. Walking a normal day without symptoms. Single-leg loading and hopping tolerated. And a strength measure you have actually tested, with the number written down.

Hopping is in that list and in two rows of the table below, so the clearance check belongs here rather than further down the page. Before you ask anyone to hop, screen for osteoporosis or a fall in the past year, a hip or knee replacement, pregnancy or the first year after birth, urinary leaking with jumping or coughing, uncontrolled blood pressure or a heart condition. Each of those is covered later on this page, and the wording to hand to the patient sits on ACL return to running.

Everything past that belongs to the tissue, and the gates come from sources of very different strength.

Tissue	The gate before a first run	Source	What the source is
Low-risk bone stress injury of the tibial or metatarsal diaphysis, meaning the posterior tibial cortex and the dorsal metatarsal cortex	Pain free during daily activities for at least 5 consecutive days, and able to perform consecutive hops without pain	Warden and colleagues, 2021	Clinical commentary reasoning from loading science
Tibial bone stress injury, more broadly	Bony tenderness resolved, pain-free walking, radiological healing where the site is high risk, strength and loading tests passed, contributing factors identified	George and colleagues, 2024	Scoping review of 50 studies, recommendations rated level IV
Achilles tendinopathy	Symptoms held inside the pain monitoring model's limit during loading and the next morning, and not creeping up week to week	Silbernagel and colleagues, 2007	Randomized controlled trial in 38 patients, 19 of whom kept running and jumping under a pain monitoring model while 19 stopped those activities for 6 weeks
After ACL reconstruction	Full extension, flexion at least 95% of the other side, no more than a trace of effusion, quadriceps strength and eccentric impulse in a countermovement jump each over 80% symmetry, pain-free repeated single-leg hopping (the guideline calls them pogos), pain-free jogging in water and on an anti-gravity treadmill	Kotsifaki and colleagues, 2023	Clinical practice guideline, criteria proposed in the absence of research findings
Hamstring strain	No agreed gate for a first run. The studies set criteria for return to play, and those leaned on pain, clinical assessment and performance tests	Hickey and colleagues, 2017	Systematic review of 9 studies, 601 acute injuries, return to play rather than return to running
Postnatal runner	Low impact only for the first 3 months, then a return between 3 and 6 months at the earliest, after a pelvic health assessment and load and strength testing	Goom and colleagues, 2019	Guideline whose timing and testing recommendations the authors grade level 4, expert opinion

Read the right-hand column before you borrow a number from it. The ACL criteria are the most completely specified set anyone has published and the guideline group says plainly it proposed them despite an absence of research findings. The tibial review rates its own recommendations level IV. The postnatal timings are expert opinion by the authors' own account.

None of that makes them worthless. It makes them agreements rather than findings, and an agreement you have written into a note is still better than a date.

The mistake we see most is transplanting a criterion across tissues. An 80% quadriceps index belongs to a reconstructed knee. It tells you nothing useful about a healing tibia, where the question is bone tenderness and imaging, and nothing about a postnatal pelvic floor.

Every row above was drawn from adult populations. A growing runner is a separate assessment, because a bone stress injury near an open growth plate is a medical and imaging question before it is a loading one.

Does the evidence support any threshold?

Not yet, for any tissue, against a running outcome. That is the honest position. It should change how you talk about your own protocol, not make you abandon it.

The clearest measurement of this is in ACL. A scoping review collected the criteria used for return to running across 201 studies and found 205 time-based criteria, with a median of 12 weeks, a range from 5 to 39 weeks, and fewer than one in five of those studies adding any clinical, strength or performance criterion at all (Rambaud and colleagues, 2018). The Aspetar guideline adds that there are no conclusive results on whether running at or before 12 weeks is safe, and that the prospective work which would answer it has not been done (Kotsifaki and colleagues, 2023).

Away from the knee the picture is thinner again. George and colleagues (2024) screened 50 studies on tibial bone stress injury and found 39 of them were reviews or clinical commentaries, with 2 randomized trials in the whole set. Hickey and colleagues (2017) looked at 9 studies covering 601 acute hamstring injuries and concluded that the field relies on the patient's own perception of pain and on subjective clinical assessment, and called for criteria that are both more objective and practical in a clinic.

So you are choosing between a measured criterion whose predictive value is unproven and a date whose predictive value is zero. Take the criterion, say out loud that it is an agreement, and write down the number you measured. The Bern consensus frames the whole decision as a shared one between clinician, athlete and anyone else with a stake in it, which is the honest shape when the evidence cannot settle it for you (Ardern and colleagues, 2016).

Building the walk-run progression

Two runners get nothing from this section until someone else has signed it off. One is a bone stress injury at a high-risk site, which Warden and colleagues (2021) name as the superior cortex of the femoral neck and the anterior cortex of the tibia. The numbers below come from an approach that paper says does not apply there, so the loading plan belongs to the team managing the bone. The other is a runner with a hip or knee replacement, because jogging sits among the activities the surgeon consensus this site follows does not recommend after a hip or knee replacement (Healy and colleagues, 2001), which makes a walk-run the implanting surgeon's decision before it is yours. Both are set out in full further down, under the bone criteria and in the strength section.

Volume before speed, every time. Both the tibial review and the loading commentary say it in those terms, and it is the one piece of progression advice the sources agree on (George and colleagues, 2024; Warden and colleagues, 2021).

The shape most clinics use is a walk-run: an easy walk with short bursts of running inside it, the running time built across sessions rather than within one. As a general range, first sessions sit at a few minutes of actual running, every other day and not daily, on flat even ground, with no hills, no camber and no turns at speed. Those are starting numbers to adjust to the tissue, the history and the sport, not a protocol to hand over unchanged.

For bone the sources are more specific than that, and the detail is worth copying exactly. Warden and colleagues (2021) start with 30 minutes of dedicated pain-free walking, then substitute increasing time increments of jogging at about half the runner's usual pace before any running, and they ask you to treat each running bout as a controlled bone-loading session rather than a conditioning one. Perceived exertion is therefore the wrong guide here, because cardiovascular and muscular capacity recover faster than bone does, so a session that feels easy can still be too much. And when you add high-speed work, drop the running volume while you do it.

Change one variable per session, so that a bad morning tells you what caused it. And judge by symptom provocation during loading, after it and the day after, which is the window Warden and colleagues use. Those two rules are what keep a progression honest.

Nielsen and colleagues (2014) followed 874 healthy novice runners and found no difference in overall injury rates between those who progressed their weekly distance, over a 2-week period, by less than 10%, by 10% to 30% and by more than 30%. Only the distance-related injuries were raised in the fastest group, and that association did not reach significance (hazard ratio 1.59, 95% confidence interval 0.96 to 2.66). Check which injuries made up that raised category before you carry the 30% figure onto a bone case: patellofemoral pain, iliotibial band syndrome, medial tibial stress syndrome, patellar tendinopathy, gluteus medius injury, greater trochanteric bursitis and injury to the tensor fascia latae. Tibial stress fracture sat in the other grouping, where the rate fell rather than rose as progression increased. It is a cohort of novices rather than of runners coming back from an injury, so read it as a reason to progress the specific tissue and not as support for a generic percentage.

There is a rate for a runner coming back from bone, and it is tighter than anything in that cohort. Warden and colleagues (2021) suggest a gradual increase in workload of about 5% to 10% a week in run time, alongside close monitoring for any symptom provocation, and they say in the same breath that validated measures for identifying spikes in workload are still in development and that neither clinician nor athlete should rely on a single metric such as the acute to chronic workload ratio. Read the percentage as a brake rather than a permission.

Count the whole day and not just the run. A typical return-to-run program adds 4000 to 7000 loading cycles, meaning steps, within a day, which is why the same paper asks you to combine run time and speed with step counts across the entire day (Warden and colleagues, 2021). A runner who jogs 8 minutes and then walks a city and a dog has loaded that bone well past what the plan on paper says. Ask about it at every review. The runner can read it off a phone without any equipment from you.

Exercise progression covers how to move a single variable without losing the thread of the program.

Criteria for a bone stress injury, a tendon, an ACL and a postnatal runner

Bone first, because it is the one where getting it wrong costs the most. Warden and colleagues (2021) report that injuries at these low-risk sites account for more than half of all bone stress injuries, describe optimal load as symptom driven with no symptoms during, after or the day after loading, and say a return-to-run program can be initiated once the athlete is pain free during daily activities for at least 5 consecutive days. Their own summary table pairs that with a second item, consecutive hops without pain, so the gate is both and not just the days. They note that an initial period of partial or non-weight bearing is typically needed.

Read that gate off the paper and not off a protocol sheet. The abstract and the summary table both say 5 consecutive days while the body text says at least 5 consecutive days, and the body governs, as it does for the strengthening dates further down. Five days is the floor, not the appointment. And neither source puts a number on the hops: the paper says consecutive hops without pain and the ACL guideline says repeated single-leg hopping, so pick the number yourself, write down the number you used, and give the runner the rule that any pain at the site stops the test rather than a count to reach.

The pain number matters here more than anywhere, because this is where a clinician who mostly treats tendons goes wrong. The paper's own figure sets the only safe and acceptable pain level for a bone stress injury at 0 out of 10 during loading, after it, and the day following it, and contrasts that with the 2 to 3 out of 10 accepted while loading soft tissue. Any pain, with activity or at rest, means the pathology has been irritated and the load has to be modified. On a bone case there is no allowance to spend.

George and colleagues (2024) add the item that decides the high-risk sites: radiological healing, before running, where the site warrants it. Shin splints covers the medial tibial stress syndrome end of this picture, which is a different problem from a frank stress fracture.

Which bone, and which surface of that bone, decides whether the first row of that table applies at all. Warden and colleagues (2021) say in their own text that they are addressing low-risk injuries of the tibial and metatarsal diaphyses, which sit on compressive surfaces, and that their concepts are not applicable to high-risk injuries, which tend to sit on tensile surfaces, carry a higher risk of the pathology progressing and need more definitive management. The two they name are the superior cortex of the femoral neck and the anterior cortex of the tibia. An anterior cortex tibial injury is therefore not that first row with a longer timeline attached to it. It is a different pathway, and reading it as a low-risk tibia is the error that ends in a complete fracture.

In the foot, the American Academy of Orthopaedic Surgeons flags three sites that behave differently from a shaft injury: the sesamoids, where the diagnosis is hard to make on plain films, the tarsal navicular, where the pain is difficult to localize, and the base of the fifth metatarsal, where the blood supply is poor, healing takes longer and is less predictable, nonsurgical management usually means at least 6 weeks of immobilization without weight bearing, return to sport is usually not before 12 weeks, and surgery is sometimes offered to stabilize the bone. Treat anything that is not a tibial or metatarsal shaft injury as potentially high risk until the imaging and the team managing the bone say otherwise, and route those sites to an orthopaedic opinion and not into a walk-run.

A normal early image is not the reassurance it looks like. That same source notes a stress fracture may not be visible on the first X-ray and may only appear weeks later, as callus forms around the site, which is also when the fracture line often becomes visible, and that where a stress fracture is suspected but not seen, an MRI or a bone scan is the next test. A negative X-ray does not clear a runner to load. A high-risk site with a normal plain film is an unanswered question rather than a green light. And where a bone stress injury has been fixed surgically, the loading plan belongs to the operating surgeon and nothing on this page overrides it.

One item that decides whether there is a next injury is not in the leg at all. Warden and colleagues (2021) ask you to screen for relative energy deficiency in sport, using the RED-S clinical assessment tool, and they list a low body mass index, delayed menarche and absent or irregular periods among the established risk factors in female runners. A previous bone stress injury is the one they call the most easily recognized and universal risk factor for the next. Two things they report, from work they cite rather than from their own data, change where you point the program: disuse after a bone stress injury causes bone loss in both legs, at its greatest around the 12 week mark, which is when a runner is usually being progressed back, and repeat injuries after a return to training turn up more often in the uninjured leg. So the strength and jump work goes on both sides, and at review you ask about the uninjured leg as well.

Warden and colleagues also limit analgesics and nonsteroidal anti-inflammatory drugs to the short term, which they put at sooner than 7 days, and only for resting pain and night pain. Medication is not ours to set. The point for a loading plan is that pain relief which lets a runner run has removed the signal the plan depends on.

Tendon runs the other way. The gate is not absence of pain but a pain limit the patient can hold, during the session and the next morning, which is the model Silbernagel and colleagues (2007) tested in Achilles tendinopathy while people kept training. Read that trial for what it showed: 19 of the 38 kept running and jumping under the model while 19 stopped those activities for 6 weeks, and the two groups improved by similar amounts, so the finding is that continued loading did no harm rather than that it did more good. The model itself has numbers, and they belong in the patient's instructions, not in a general sentence about pain: up to 5 out of 10 during and just after loading, back to their usual level by the next morning, and not creeping up from one week to the next. Achilles tendinopathy covers the condition and the pain monitoring model covers the scale and the next-morning check you give the patient.

ACL has the criteria set out in the table above, plus one item not in any guideline: the surgeon's permission, which outranks the list. A meniscal or cartilage repair in the same operation often pushes impact later, and how much later is not a decision a protocol makes. The guideline takes the same line: it reserves early accelerated rehabilitation for isolated reconstructions and says that where concomitant meniscal or cartilage injury is present, the early phase is adapted according to the surgeon's instructions (Kotsifaki and colleagues, 2023).

Note where its criteria came from as well, because the review behind them excluded trials reporting only on concomitant injuries such as other knee ligament, meniscal or cartilage injury. The set was built around an isolated reconstruction, so a knee that also holds a meniscal repair, a cartilage procedure or a second ligament is being measured against criteria nobody tested on it, and the date comes from the operating team.

Read two of the guideline's items closely before you record a pass. The jump criterion is eccentric impulse during a countermovement jump, not jump height. And the guideline lists aqua jogging and anti-gravity treadmill running together, so a clinic with neither has an item it has not tested and not an item the patient passed.

The guideline also puts a scope on its own list. It defines the running it is setting criteria for as running with a volume and intensity that achieves cardiovascular adaptation, and it offers no separate, lower set for a first jogging interval. That gap is not a licence to start someone earlier against an untested standard, because there is no other published set to measure them against.

Hamstring is where the literature is least help and where the clinic has to supply the structure. Hickey and colleagues (2017) reported the lowest mean times to return to play in studies that used isokinetic dynamometry in the decision, 12 to 25 days, and the lowest reinjury rates, 1.3% to 3.6%, in studies using the Askling H-test. Those are associations across heterogeneous studies, not a protocol, so use them as a reason to put an objective test in the decision and not as thresholds. Hamstring strain covers the rehabilitation that sits underneath.

The postnatal runner needs a different assessment, not a modified ACL one. Goom and colleagues (2019) advise low-impact work for the first 3 months, a return to running between 3 and 6 months at the earliest, a pelvic health assessment first, and load and strength testing before impact. Leaking, heaviness, dragging or pelvic pain during or after exercise are signs the runner is not ready to progress, and so is vaginal bleeding unrelated to a period during or after an attempt at impact, which the guideline sends back to her own care provider. Postnatal core and pelvic floor carries that pathway for the patient.

The load and strength testing in that guideline is specified, so run it as written. The load and impact assessment asks for walking 30 minutes, single leg balance 10 seconds, single leg squat 10 repetitions each side, jogging on the spot 1 minute, forward bounds 10 repetitions, hopping in place 10 repetitions each leg, and a single leg running man, opposite arm and hip flexion and extension with a bent knee, 10 repetitions each side, all of it without pain, heaviness, dragging or incontinence. The strength testing counts repetitions to fatigue, aiming for 20 of each, on the single leg calf raise, single leg bridge, single leg sit to stand and side lying abduction. Read the authors' own note on that second list before you use it as a gate: weakness there is not a barrier to returning to running, it is a map of where the strength work goes (Goom and colleagues, 2019). The pelvic floor findings are the part that holds a runner back, not the repetition counts.

How do you progress strength alongside the running?

You keep it, and you resist the pull to swap it for mileage. The strength work is what earned the run and it is the first thing a motivated runner drops.

Keep calf and plantarflexor capacity in the program throughout, because it is the tissue doing most of the work in a stride and the first to fall behind. Single leg heel raises are the usual measure and the usual exercise at once. Plyometric exposure sits between strength and running, not after it, and pogo hops are the low-amplitude starting point most patients tolerate.

One tissue is the exception, and it cuts against that rule. Warden and colleagues (2021) hold resisted weight-bearing plantar flexion back, weighted calf raises being their own example, until bone tenderness has resolved. That instruction is written for a metatarsal injury, where the shaft takes large bending moments, while the tibia is governed instead by a schedule tied to the MRI grade and to which surface of the bone is injured rather than to how the session feels. Applying the tenderness rule to both is this site's conservative default rather than the paper's wording, and it is worth knowing which is which, because the two rules answer different questions.

Their summary table gives weeks 3 to 4 depending on grade, and their text says what that means: standing heel raises, seated soleus raises and leg presses may be started in weeks 3 to 4 for a Fredericson grade 2 posteromedial tibial injury, provided the patient can walk and perform them without pain, but loaded resistive exercise waits until weeks 5 to 6 for grade 3 to 4a and until week 8 for grade 4b. Plantar flexor work is delayed again where the injury is on the anterior cortex, because that is the high-risk tensile surface, and it may start earlier on the posteromedial tibia because the posterior tensile stresses there are low. Hamstring curls wait a further 2 weeks beyond each of those dates, since the resistance sits distal to the site and bends the tibia in tension, while hamstring, quadriceps and hip strengthening can start earlier if the resistance is applied proximal to the injury and the exercise is pain free. So weeks 3 to 4 is the grade 2 figure and not the general one. On a bone case the calf work starts later and starts off the floor, and a loaded heel raise is not the week-one exercise it is for a tendon.

Most runners turn up without a grade, so settle in advance what you do then, because a blank here gets filled with a guess. The grade in that schedule is an MRI grade, first described by Fredericson and colleagues (1995) in 14 runners with 18 symptomatic legs and medial tibial pain, and those authors recommended MRI over a bone scan for grading a tibial stress lesion. Neither a plain film nor an examination produces a Fredericson grade.

With no report in front of you the earliest date in the schedule does not apply, because nothing you can see rules out a grade 4b, so the working default is the latest date in it: week 8 for loaded resistive plantar flexor work and week 10 for hamstring curls, brought forward only when a report puts the injury lower. Two things govern either way and need no imaging at all, which are bone tenderness resolved before any resisted weight-bearing plantar flexion, and 0 out of 10 during loading, after it and the day following. Some findings argue for imaging rather than for a week number: Fredericson and colleagues found pain with daily walking, tenderness localized to the tibia and pain on direct or indirect percussion went with the more severe injuries, and tenderness over the anterior tibial crest points at a high-risk site rather than at a worse low-risk one. Where the site or the grade cannot be established, the loading plan is a question for the team managing the bone, and writing that in the note is better than inventing a date.

Where the dates do allow it, Warden and colleagues give the dose the strength work is heading for: high load at 60% to 80% of the 1-repetition maximum and low repetitions, 3 to 6 sets of 5 to 15. On a metatarsal injury a good deal can start early instead, being foot intrinsic work such as towel curls and non-weight-bearing short foot exercises, leg presses pressing through the heel, and knee and multiplanar hip strengthening. Keep ankle dorsiflexion range alongside it. Dorsiflexion falls away quickly once normal walking is lost, and a stiff ankle brings the heel up early in stance and loads the metatarsal harder on the first run back, which is the opposite of what the plan is for.

Before you prescribe any hopping, check the list that needs clearance first: osteoporosis or a fall in the past year, a hip or knee replacement, pregnancy or the first year after birth, urinary leaking with jumping or coughing, uncontrolled blood pressure or a heart condition. The patient-facing wording for each of those is on ACL return to running and is the version to hand over, because it is the site's recorded wording and it says what to do rather than only who should ask.

Give the stop rule with the clearance rule, because the clearance half is the half that gets given on its own. Stop the hopping and reassess if urine leaks while the patient is doing it, or if they feel pressure or heaviness low in the pelvis.

A hip or knee replacement is not a box to clear and move on from: the surgeon consensus this site follows advises against jumping and hopping after a joint replacement long term, not only for the first months, so the go-ahead comes from the surgeon and the honest default is that impact stays out.

That restriction covers the running and not only the hopping, because the clearance list sits under hopping while the thing you are about to prescribe is a walk-run. Jogging sits among the high-impact activities the same consensus does not recommend after a hip or knee replacement (Healy and colleagues, 2001), so a walk-run for a replacement patient is the surgeon's decision before it is yours. Those consensus recommendations were drawn from surveys of hip, knee, shoulder and elbow surgeons, so an ankle replacement or any other lower limb implant is not covered by them and is a question for the operating surgeon rather than an extrapolation from the knee.

Say what that advice rests on, because this is one of the places surgeons genuinely differ. In the survey of the American Association for Hip and Knee Surgeons, more than 95% of responses placed no limit on level walking, stair climbing, level cycling, swimming or golf, higher-impact activities were more commonly discouraged but with considerable variability between surgeons, recommendations after a hip replacement were more liberal than after a knee, higher-volume surgeons tended to be more liberal, and not one respondent said there was strong scientific evidence behind the recommendation they gave (Swanson and colleagues, 2009). It is a community standard rather than a survivorship finding, the long-term effect of higher-impact activity on an implant is still unresolved, and none of that moves the default. Impact stays out unless the surgeon who fitted the implant says otherwise.

And after a hamstring or a quadriceps tendon graft the hopping items wait until at least 16 weeks after surgery unless the operating team has cleared them earlier, because the tissue still rebuilding is the harvest site rather than the graft in the knee (Briem and colleagues, 2026). That can land later than the 12 week mark most ACL protocols use for running, which is one more reason the first run is cleared by the team that operated and not by a list.

The 16 weeks deserves the same reading as the replacement advice. Briem and colleagues charted what guideline documents recommend rather than how a harvest site heals, and the review is level V. In its graft-specific table the delay comes from two documents for the semitendinosus harvest, which hold plyometrics back across weeks 0 to 16, and from one document for the quadriceps tendon harvest, which begins plyometrics from week 16 while avoiding sudden spikes. The authors' own account of why those two sites are singled out is that taking the patellar or quadriceps tendon leaves the tendon's continuum intact, while taking the semitendinosus has a direct mechanical effect on hamstring function, because the tendon atrophies and retracts.

A bone-patellar tendon-bone harvest carries no plyometric delay anywhere in that table. Its charted restrictions are early ones: protecting the extensor mechanism and anterior knee pain over the first 8 weeks, holding isometrics at 45 to 60 degrees for patellofemoral comfort, progressive patellar mobilization, and a slower return of quadriceps strength throughout. Bone-to-bone healing at both ends of the graft is the usual reason given for allowing a faster progression.

Do not read that absence as a clearance for impact. The patellar tendon defect and the two bone plugs are a harvest site as well, donor-site morbidity is highest with this graft, and anterior knee and kneeling pain is reported in up to 72% of these patients against up to 44% after a hamstring graft and up to 9.3% after a quadriceps tendon graft. Patellar fracture after a graft taken with bone blocks is reported between 0.1% and 2%, or as high as 8.8% when occult fractures are counted, and rupture of the patellar or quadriceps tendon at around 1% (Runer and colleagues, 2023). No published date tells you when that extensor mechanism will take hopping, which is exactly why the date comes from the operating surgeon. The one move to avoid is inferring an earlier date from the silence.

Where the runner is heading back to a cutting or pivoting sport and not to road running, the first run is a staging post inside a longer process, and the return to sport test battery covers the measurement that comes later.

What should go in the note, and what do you tell the patient?

Six things in the note, and the first one is the one audits find missing. Which criterion you used and whose it is. The measured value, the side, how you measured it, and the date. The first session you prescribed, in minutes and frequency. The morning-after rule you gave, and the review date.

Add the surgical, obstetric or medical clearance where one applies, named and dated, because that is the item a later reader cannot reconstruct.

Writing the source of the criterion matters more than it sounds. A note that says "met return to running criteria" is unreadable in six months. A note that says "Aspetar first-run criteria, quadriceps index 84% on handheld dynamometry, flexion 96%, no effusion, 3 of 3 single-leg hops pain free" tells the next clinician what was actually true. Outcome measures in physical therapy covers choosing the measure and reading the change in it.

What you tell the patient is shorter, but it needs three rules rather than one, because two of the three happen when you are not in the room. During the run: pain at the injured site ends that run, so walk the rest of the way home instead of finishing the session, and on a bone injury that means any pain at all, at any level. The next morning: if the limb is more swollen or more painful on waking than the day before, go back to the last session that settled and build from there, and two worse mornings in a row means contacting you rather than making a third attempt, which is usually a dose to cut rather than a stop. And the short list of things that end the plan and get a call the same day rather than a smaller session, which is the list in the next section.

Say who gives what, because that is where a hand-off fails. The walk-run, the during-the-run rule and the morning-after rule come from you. The surgical, obstetric or medical clearance comes from the team holding it, and a runner has not had that conversation merely because you wrote the plan around it. Then hand over the return to running plan, or ACL return to running after a reconstruction, and not this page. Those carry the stop signs and the precautions in the patient's own register, and they carry them above the exercises instead of below them.

When do you hold a runner back?

Six situations, and none of them is a judgement call you should make quickly.

Swelling above a trace, because the joint is still reacting to what it has already had. A criterion not met, including the one you cannot test because you have no pool or anti-gravity treadmill, which is a reason to build impact another way and not to count the item as passed. A surgical restriction that has not been cleared by the team that operated. A high-risk bone stress injury site without the imaging. Pelvic floor symptoms during or after loading in a postnatal runner. And two consecutive worse mornings, which is a dosing answer rather than a stop.

Four presentations are not dosing problems and do not belong on that list, because they leave the loading conversation altogether. A reconstructed or repaired knee that locks, catches or gives way again, or that pops, shifts or swells quickly after a twist, a kneel or a deep bend, which is how a graft or a meniscal repair announces that it has failed. A joint that is hot, red, tender or swollen after any joint operation, or a wound that is oozing or opening, or a temperature alongside either, which is a possible infection in the wound or the joint and not a reaction to a session. Sudden pain at the back of the ankle or heel with loss of push off, which is a possible Achilles rupture and goes to urgent care or an emergency department the same day even if the patient can still walk. And a bone stress injury whose pain has moved into rest or into the night, or where the tenderness sits on the anterior tibial crest, which needs re-imaging and a surgical opinion before any further loading rather than a smaller session. The first two go to the operating team the same day, and the patient-facing wording for all of them is on the condition pages rather than here: ACL reconstruction rehab, meniscus tear and Achilles tendinopathy.

There is a seventh, and it is not a failure of the protocol. A runner who has met everything and does not want to run yet. The Bern consensus puts the decision in shared hands, and a runner who is not ready is part of that decision, not an obstacle to it (Ardern and colleagues, 2016).

The short version

Pick the criteria by tissue, not by borrowing the set you know best. Say out loud which of your criteria are agreements rather than findings, because most of them are. Build volume before speed, change one variable at a time, and judge by the next morning. Then write down what you measured, how, and under whose criteria, because that sentence is the whole value of having a protocol in the first place.

References

1. George ERM, Sheerin KR, Reid D. Criteria and Guidelines for Returning to Running Following a Tibial Bone Stress Injury: A Scoping Review. Sports Medicine. 2024;54(9):2247-2265. https://doi.org/10.1007/s40279-024-02051-y
2. Warden SJ, Edwards WB, Willy RW. Optimal Load for Managing Low-Risk Tibial and Metatarsal Bone Stress Injuries in Runners: The Science Behind the Clinical Reasoning. Journal of Orthopaedic and Sports Physical Therapy. 2021;51(7):322-330. https://doi.org/10.2519/jospt.2021.9982
3. Fredericson M, Bergman AG, Hoffman KL, Dillingham MS. Tibial stress reaction in runners. Correlation of clinical symptoms and scintigraphy with a new magnetic resonance imaging grading system. American Journal of Sports Medicine. 1995;23(4):472-481. https://doi.org/10.1177/036354659502300418
4. Rambaud AJM, Ardern CL, Thoreux P, Regnaux JP, Edouard P. Criteria for return to running after anterior cruciate ligament reconstruction: a scoping review. British Journal of Sports Medicine. 2018;52(22):1437-1444. https://doi.org/10.1136/bjsports-2017-098602
5. Kotsifaki R, Korakakis V, King E, et al. Aspetar clinical practice guideline on rehabilitation after anterior cruciate ligament reconstruction. British Journal of Sports Medicine. 2023;57(9):500-514. https://doi.org/10.1136/bjsports-2022-106158
6. Hickey JT, Timmins RG, Maniar N, Williams MD, Opar DA. Criteria for Progressing Rehabilitation and Determining Return-to-Play Clearance Following Hamstring Strain Injury: A Systematic Review. Sports Medicine. 2017;47(7):1375-1387. https://doi.org/10.1007/s40279-016-0667-x
7. Goom T, Donnelly G, Brockwell E. Returning to running postnatal: guidelines for medical, health and fitness professionals managing this population. March 2019. https://absolute.physio/wp-content/uploads/2019/09/returning-to-running-postnatal-guidelines.pdf
8. Silbernagel KG, Thomeé R, Eriksson BI, Karlsson J. Continued sports activity, using a pain-monitoring model, during rehabilitation in patients with Achilles tendinopathy: a randomized controlled study. American Journal of Sports Medicine. 2007;35(6):897-906. https://doi.org/10.1177/0363546506298279
9. Nielsen RØ, Parner ET, Nohr EA, Sørensen H, Lind M, Rasmussen S. Excessive progression in weekly running distance and risk of running-related injuries: an association which varies according to type of injury. Journal of Orthopaedic and Sports Physical Therapy. 2014;44(10):739-747. https://doi.org/10.2519/jospt.2014.5164
10. Ardern CL, Glasgow P, Schneiders A, et al. 2016 consensus statement on return to sport from the First World Congress in Sports Physical Therapy, Bern. British Journal of Sports Medicine. 2016;50(14):853-864. https://doi.org/10.1136/bjsports-2016-096278
11. Briem K, Zebis MK, Haraldsson BÞ, Bencke J, Fernandes L. Rehabilitation guidelines after autograft anterior cruciate ligament reconstruction need more graft-specific exercise recommendations: A scoping review. Knee Surgery, Sports Traumatology, Arthroscopy. 2026;34(1):83-101. https://doi.org/10.1002/ksa.12666
12. Runer A, Keeling L, Wagala N, et al. Current trends in graft choice for primary anterior cruciate ligament reconstruction, part II: in-vivo kinematics, patient reported outcomes, re-rupture rates, strength recovery, return to sports and complications. Journal of Experimental Orthopaedics. 2023;10:40. https://doi.org/10.1186/s40634-023-00601-3
13. Healy WL, Iorio R, Lemos MJ. Athletic activity after joint replacement. American Journal of Sports Medicine. 2001;29(3):377-388. https://doi.org/10.1177/03635465010290032301
14. Swanson EA, Schmalzried TP, Dorey FJ. Activity recommendations after total hip and knee arthroplasty: a survey of the American Association for Hip and Knee Surgeons. Journal of Arthroplasty. 2009;24(6 Suppl):120-126. https://doi.org/10.1016/j.arth.2009.05.014
15. American Academy of Orthopaedic Surgeons. Stress fractures of the foot and ankle. OrthoInfo. https://www.orthoinfo.org/en/diseases--conditions/stress-fractures-of-the-foot-and-ankle/

Written and checked by the PocketPhysio editorial team. Last updated 2026-10-24.