Research · Sprint Mechanics

“Drive Your Knees,” “Claw the Ground”: What Sprint Biomechanics Research Actually Says

By Muhammad Dzulhisham · MSc Sports Coaching, NSCA-CSCS

Sprint coaching cues travel by word of mouth more than almost anything else in strength and conditioning. “Drive your knees.” “Claw the ground.” “Pump your arms like you're punching someone.” Most of these phrases were coined by sprinters and coaches decades before force plates, high-speed video, and motion-capture systems existed to check them against anything. Over the last twenty-five years, sport biomechanics has caught up — measuring exactly what happens at the hip, knee, ankle, and ground during a sprint stride, and in several cases testing the cues themselves against sprint time. The findings don't sort neatly into “coaches were right” or “coaches were wrong.” Some inherited wisdom holds up well. One widely taught safety cue has essentially no evidence behind it. And a separate body of research suggests a biomechanically accurate cue can still underperform simply because of how it's worded.

What actually separates a fast sprinter from a slow one

The starting point for almost all of this research is a 2000 study by Peter Weyand and colleagues in the Journal of Applied Physiology, comparing 33 runners of very different sprinting ability at their own top speeds on a treadmill.¹ The intuitive assumption going in was that faster people simply cycle their legs through the air more quickly — get the recovery leg back to the ground sooner, take more steps per second. The data said the opposite. Limb repositioning speed barely differed between fast and slow runners. What differed was how much force each runner applied to the ground, relative to body weight, during the brief window their foot was in contact with it. Faster sprinters hit the ground harder in the same or even shorter contact time — they weren't moving their limbs faster in the air, they were producing more force against the ground while they were on it.

That single finding reframes a lot of sprint coaching. Cues built around leg speed or turnover rate — “move your legs faster,” “quick feet” — are targeting a variable the evidence says isn't what actually discriminates sprint speed. The trainable lever is force production into the ground during an extremely short window, not the speed of the swing.

“Push the ground away,” not “claw it back”

Force production alone isn't the whole story, either. A 2011 study by Jean-Benoît Morin and colleagues in Medicine & Science in Sports & Exercise introduced what's now called the “ratio of force” — the proportion of total ground reaction force that's actually oriented horizontally, in the direction of travel, versus wasted vertically.² Testing sprinters on an instrumented treadmill and over 100m on a track, the researchers found that the technical ability to orient force forward was a stronger predictor of sprint performance than total force production on its own. Two athletes can produce identical total force into the ground and still sprint at different speeds, because one of them is directing more of that force horizontally.

This gives some real backing to acceleration cues built around pushing the ground behind and away from the body — the forward-lean, positive shin-angle coaching common in the first few steps out of a start — over cues that emphasize stomping or driving straight down. The direction force is applied matters as much as how much of it there is.

The cue with essentially no evidence behind it: “claw the ground”

“Claw the ground”, “active pawback”, and “back-side mechanics” cues ask a sprinter to actively accelerate the foot backward and downward just before and at ground contact, on the theory that this reduces braking forces and protects the hamstrings from injury. It's one of the most widely taught cues in sprint coaching, and it's usually justified specifically as an injury-prevention strategy.

A 2024 review in Sports Medicine by Bramah, Mendiguchia, Dos'Santos, and Morin looked directly at where hamstring strain injuries actually occur in the sprint stride and what the biomechanics literature says about back-side mechanics as a protective factor.³ The injury risk, the review confirms, really does concentrate in a specific window — terminal swing, just before the foot lands, where hamstring muscle-tendon forces reach up to ten times body weight and the muscle-tendon unit lengthens by roughly 10% while contracting eccentrically to decelerate the swinging shank. That part of the picture is well established.

But the specific claim that cueing “back-kicking” or active pawback protects against that injury doesn't hold up. The review notes that coaches rate back-side mechanics and trailing-leg kick-back as “highly important” for injury prevention, yet the actual prospective evidence — studies that tracked sprinters' kinematics and then recorded who later got injured — found no kinematic differences between the athletes who went on to sustain a hamstring strain and those who stayed healthy. The authors describe the link between back-side mechanics and hamstring injury as “theoretical” at best.

Based on current evidence, there appear to be only theoretical associations between back-side mechanics and hamstring strain injury.

That's a meaningfully different message from what gets taught in most sprint programs. The mechanics of the injury are real and well mapped. The specific coaching fix built around them isn't supported by the prospective data that's actually been collected on it.

Is “drive your knees” a cause, or a downstream effect?

Knee drive is trickier to sort out, because two separate lines of research point in slightly different directions depending on which part of the stride they're looking at.

A 2012 study by Dorn, Schache, and Pandy in the Journal of Experimental Biology modeled how the relative contribution of different muscle groups to forward propulsion changes as running speed increases.⁴ Below roughly 7 metres per second — jogging and moderate running — the ankle plantarflexors do most of the propulsive work. Above that speed, which covers essentially all competitive sprinting, the strategy shifts: hip muscles — the iliopsoas, glutes, and hamstrings — become the dominant driver of the increases in stride frequency that produce further speed gains, because the ankle muscles are shortening too fast at sprint speeds to keep generating high force. In other words, vigorous, forceful hip flexion and extension is a genuine, measurable feature that separates true sprinting from jogging — not an aesthetic add-on. The visible “knee drive” that comes with that hip action reflects a real, sprint-specific shift in muscle recruitment.

Where it gets more complicated is what's actually generating that motion. A 2008 study by Bezodis, Kerwin, and Salo in Medicine & Science in Sports & Exercise analyzed joint kinetics during the ground-contact phase of maximum-velocity sprinting in well-trained sprinters.⁵ It found that the knee joint itself contributes little to power generation during the latter part of stance — the major power-generating events are hip extensor power early in stance and ankle plantarflexor power late in stance. The knee, at the point of ground contact where force actually gets applied, isn't where the engine is.

Put together, these two findings support a specific, fairly narrow claim: forceful hip action at high speed is real and worth training, but the knee joint isn't the power source that produces it. A cue that sends an athlete's attention to lifting the knee itself — a single joint, mid-air, during the recovery phase — is arguably pointing at the visible result of hip-driven mechanics rather than the mechanism that creates it.

Acceleration and max velocity are mechanically different problems

A separate reason a single cue set struggles to work across an entire sprint is that acceleration and top-speed running are not the same movement. A 2014 study by Nagahara and colleagues in Biology Open tracked step-by-step kinematics across a full acceleration phase and identified distinct transition points within it.⁶ Early in acceleration, the foot contacts at or behind the body's centre of mass and the knee stays extended through stance. At a measurable transition point later in the acceleration phase, that changes: the foot begins contacting ahead of the centre of mass, and the knee starts flexing during stance — kinematic signatures of the shift toward upright, top-speed running. Step length keeps increasing and support time keeps shortening across the whole phase, but the underlying joint mechanics are not one continuous pattern; they change qualitatively partway through.

That matters directly for coaching cues. A forward-lean, extended-knee-at-contact pattern that's correct in the first few steps out of a start becomes the wrong target once an athlete has transitioned toward top-speed mechanics, and vice versa. Cueing “stay low and drive” through the entire length of a 40m sprint, or cueing tall, stiff-legged top-speed mechanics from the very first step, both work against a kinematic transition the research shows is actually happening in the athlete's body regardless of what they're told to do.

What the arms are actually doing

Arm action gets some of the most emphatic coaching language in sprinting — “pump,” “drive,” “punch” — on the assumption that vigorous arm swing is a direct engine of leg speed. A 2022 study by Brooks, Weyand, and Clark in Gait & Posture tested that assumption by having 17 athletes sprint 30m normally and again with their arms folded across their chest.⁷ Restricting arm motion did slow sprinters down — but only marginally: about 1.6% slower on average, less than 0.10 seconds' difference across the group. The authors concluded that the classic view of arm swing directly driving leg motion “is not well-supported” by that magnitude of effect.

That doesn't mean arm action is irrelevant. What the data suggest instead is a different job: the arms appear to function mainly as a counter-rotation mechanism, offsetting the angular momentum the legs generate so the torso doesn't twist excessively stride to stride, rather than as an independent source of forward propulsion. A cue built around active arm swing is probably improving trunk stability and rhythm more than it's adding raw speed — useful, but not the primary lever the emphatic coaching language implies.

A cue can be biomechanically correct and still fail

The last piece of this research doesn't test any specific sprint mechanic at all — it tests how a cue is phrased. A 2022 meta-analysis by Li, Zhang, Yue, Memmert, and Zhang in the International Journal of Environmental Research and Public Health pooled six studies and 166 participants comparing “external focus” cues (attention directed at the movement's effect on the environment — push the track away, explode off the ground) against “internal focus” cues (attention directed at a body part — extend your hip, drive your knee).⁸ Across the pooled data, external-focus cueing produced a small-to-moderate but statistically real advantage in sprint performance (g = 0.279, 95% CI 0.088–0.470).

The proposed mechanism is what's called the constrained action hypothesis: directing attention to a specific joint or body part interferes with the automatic, self-organizing motor control that a well-trained nervous system otherwise handles on its own, while an external focus leaves that automaticity alone. Practically, this reframes several of the cues discussed above. Even where the underlying mechanic — hip-driven knee lift, forward force orientation — is accurate, phrasing it as “drive your knee” or “extend your hip” may be a less effective delivery than the externally-framed equivalent, such as “punch the ground back” or “push the track away.” The content of the cue and the frame it's delivered in are two separate variables, and the research on attentional focus suggests the frame carries real weight of its own.

What this adds up to

None of this produces a tidy list of cues to keep and cues to throw out — the research is more specific than that, and the specificity is the useful part. Ground force magnitude and, separately, its horizontal orientation are what the data say actually differentiate sprint speed, not leg turnover. Vigorous hip-driven knee action is a real, measurable feature of true sprinting rather than an aesthetic flourish, but the knee joint itself isn't where that power is generated, which argues for cueing the hip-and-ground interaction over the knee in isolation. Acceleration and top-speed running are kinematically distinct phases with a measurable transition between them, so a single cue set applied across an entire sprint works against what the athlete's body is actually doing partway through. “Claw the ground” as an injury-prevention strategy specifically has no supporting evidence from the prospective studies that have actually tested it, even though the injury mechanics it's meant to address are real. Arm action matters more for rotational stability than direct propulsion. And separate from all of the above, how a cue is worded — pointed at a body part versus pointed at an external effect — changes how well even a mechanically accurate cue works.

Treated as a set, the findings argue for the same standard sport science applies everywhere else: check what a cue is actually claiming against what force plates, motion capture, and prospective injury data show, rather than against how long it's been repeated in a locker room.