Biomechanics and gait
Evidence: moderate
Gait is largely self-optimised, so most deliberate form changes do not help and some hurt. The clearest exception is a modest cadence increase to offload a painful joint. Footstrike has no clear winner.
Running gait is the pattern of how a runner moves: how often the feet land (cadence), how far ahead of the body they land (overstriding), how long each foot stays down (ground contact time), and which part of the foot lands first (footstrike). These are the levers people most often try to change. The evidence says most of them are already tuned close to a personal optimum, so deliberate changes rarely help and sometimes raise the energy cost or the injury risk (Hunter & Smith 2007).
Cadence and the 180 figure
The figure of 180 steps per minute comes from Jack Daniels, who counted the stride rates of distance runners at the 1984 Olympics and found nearly all of about 47 elite athletes ran at or above that rate at race pace (Daniels 1984). It was an observation of elites racing fast, not a universal target. Cadence rises with speed, so a rate seen at 5 km race pace does not apply to an easy run, and the individual optimum varies with height and pace (Marathon Handbook).
Cadence is largely self-optimised. Runners settle on a stride frequency at or near the one that minimises their oxygen cost (Hunter & Smith 2007). This is the same self-tuning that makes running economy hard to improve by conscious cueing. Anatomy explains less of the difference between runners than intuition suggests: leg length accounts for only about a fifth of the variation in cadence (Tenforde et al. 2019), so a runner’s step rate is individual rather than simply a matter of height. And across runners a higher cadence is only weakly tied to better economy, on the order of 4% of the variance, so no single cadence is efficient in general (Van Hooren et al. 2024).
When a cadence change does help
A modest rise in step rate, around 5 to 10%, lowers the load on some joints. Increasing step rate to 110% of preferred reduced peak patellofemoral joint force by about 14%, mostly through less knee flexion at landing (Lenhart et al. 2014). A 10% increase improved symptoms in runners with patellofemoral pain at four weeks and three months (Bramah et al. 2019). This is a targeted fix for a loaded joint, not a blanket prescription.
Cadence and injury risk
Offloading an already-painful joint is the clearest use, but there is also a weaker, contested signal that a habitually low cadence is itself a risk factor. In two studies of homogeneous groups — where age, training and speed vary less than in the general running population — the lowest-cadence runners fared worse: high-school cross-country runners below about 164 steps per minute had roughly seven times the rate of shin injury of those above 174 (Luedke et al. 2016), and among collegiate runners each extra step per minute was associated with about 5% lower risk of a bone stress injury (Kliethermes et al. 2021). The mechanism is plausible: at a fixed speed a higher cadence spreads the work over more, gentler loading cycles, and biomechanical modelling finds it lowers per-step load at the three most common injury sites — the patellofemoral joint, tibia and Achilles (Van Hooren et al. 2024). The caveats are real: large population studies find no cadence–injury link, cadence there is confounded with speed and fitness, and none of this proves that raising a healthy runner’s cadence prevents injury. Treat a low cadence as one modifiable factor worth addressing in a runner who keeps breaking down at the knee, shin or Achilles, not a target for everyone.
Overstriding
Overstriding means the foot lands well ahead of the body’s centre of mass, which lengthens ground contact and raises braking forces. Raising cadence shortens the stride and pulls the landing back under the hips, which is the mechanism behind the joint-offloading effect above (Lenhart et al. 2014). The useful target is the landing position, not a number on a watch.
Footstrike: no clear winner
Footstrike pattern divides runners into rearfoot, midfoot and forefoot landers. The mechanics differ. Rearfoot strikers show higher vertical loading rates; forefoot strikers land with a plantar-flexed ankle and a more flexed knee, altering loading at the knee, ankle and calf (Almeida, Davis & Lopes 2015). Neither pattern is clearly better for economy or injury. A systematic review of 53 studies found only limited, retrospective evidence linking non-rearfoot strike to lower injury rates, no economy advantage for either pattern, and no prospective trials to settle it (Anderson et al. 2020). Deliberately switching footstrike moves load rather than removing it, and can trade a knee problem for a calf or Achilles one.
Gait retraining: limited and specific
Gait retraining works best as a targeted clinical tool, not a general upgrade. The strongest case is a cadence increase to offload a specific painful joint: a meta-analysis of 19 trials (673 participants) found with moderate certainty that step-rate retraining reduces average vertical loading rate and two trials showed reduced one-year injury incidence, while footstrike and multiparameter approaches did not improve running economy (Doyle et al. 2022). Beyond that, because gait is already near a personal optimum, broad attempts to impose a ‘correct’ form tend to cost economy and provoke new strains (Hunter & Smith 2007). Footwear interacts with all of this: shoes change loading and footstrike, and barefoot running tends to push runners toward a forefoot landing, with the same trade-off of load rather than a clear net gain.
Two prospective-cohort findings add a reassurance and a caveat. First, leg, knee and ankle stiffness showed no association with injury rate over a year of follow-up, so the modest stiffness changes retraining produces are unlikely to drive injury either way (Davis & Gruber 2021). Second, a single lab or treadmill assessment represents an individual’s real-world running poorly — a runner’s in-lab gait matched only about a third of their own real-world data, and leg stiffness worst of all — so form captured once in a lab is a weak guide to how someone runs day to day (Davis et al. 2024).