The pitch for grinding a big gear at 50 rpm is seductive: you are supposedly teaching your fast-twitch fibers to run on oxygen, firing the AMPK pathway that builds mitochondria, getting a strength stimulus without leaving the bike -- while the gym rat next door only trains mTOR and bulk. It is a tidy story. It is also mostly a story. The one systematic review of the question concludes there is no strong evidence that low-cadence training beats normal cadence at matched power, the flagship study that seemed to prove it was quietly doing plyometric jumps on the side, and the peak force at your knee on the bike is about a twelfth of what a gym movement produces.
The Simple Version
Low-cadence or muscle-endurance work does one thing that is real and measurable: at a given power, it makes your fast-twitch fibers do more of the work. From there the claims outrun the evidence. It does not reliably make you faster than the same session at normal cadence, it cannot replace gym strength because the forces are an order of magnitude too small, and the neat AMPK-versus-mTOR framing is a simplification. What it genuinely is: a specificity tool for people who race at low cadence -- climbers, time-triallists, mountain bikers -- with one study showing it helps when paired with high intensity, and a real knee-loading cost that rules it out for some riders entirely.
How It Works
What Is Real and What Is Sold
The One Thing That Holds Up
The mechanical logic is sound. Power is torque times cadence, so at a fixed power, halving your cadence roughly doubles the force on each pedal stroke. More force means recruiting higher-threshold motor units -- the fast-twitch fibers that stay in reserve during easy spinning.
And there is direct evidence this happens. Ahlquist and colleagues had eight trained men cycle 30 minutes at 85% of VO2max at 50 and at 100 rpm on separate days, then took muscle biopsies. Type II fiber glycogen fell 49% at the low cadence versus 33% at the high one -- a real, significant difference -- while type I depletion was essentially the same. At matched metabolic cost, low cadence genuinely pulls the fast fibers into the work.
That is the strongest fact in favor of the method. Notice what it is not: it is not evidence that this makes you faster, or that it builds strength, or that it beats a normal session. It is evidence about which fibers get tired.
Does It Actually Make You Faster?
Here the picture falls apart. Hansen and Ronnestad reviewed the controlled trials and concluded, in plain words, that there is no strong evidence for a benefit of training at low cadence. Some studies found nothing; some found freely chosen cadence better.
The study most often waved as proof is Paton and colleagues (2009), which reported the low-cadence group improving 60-second power by 2.5%, peak power by 3.6% and lactate power by 7%. Impressive -- until you read the methods. The sessions "consisting of sets of explosive single-leg jumps alternating with sets of high-intensity cycling sprints." The low-cadence riders were doing plyometric jumps between efforts. You cannot credit the bike cadence for a result produced by a programme that also included explosive jumping.
Set against that, Kristoffersen and colleagues ran a cleaner test: 22 veteran cyclists, 12 weeks, 40-rpm intervals versus freely chosen cadence at matched work. The 40-rpm group improved nothing -- not aerobic capacity, not performance, not efficiency, not leg strength. The freely chosen cadence group raised VO2max from 58.9 to 62.2 mL/kg/min. The grinding group would have been better off spinning.
The honest summary: no guaranteed benefit at matched power, and at moderate intensity the low cadence may cost you.
The AMPK-versus-mTOR Story
The framing that low-cadence riding hits AMPK (mitochondria) while the gym hits mTOR (strength) sounds mechanistic but is doing more rhetorical work than physiological.
Two problems. First, nobody has actually measured it -- there are no human biopsy studies comparing AMPK and mTOR signalling after matched-power intervals at low versus high cadence. The claim is extrapolation. Second, and more concrete: mTOR responds to high mechanical tension, and the tension on a pedal simply is not high enough. Peak patellofemoral force on the bike tops out around 1.3 times body weight (below), whereas a gym knee-extension reaches about 12 times body weight. That is an order of magnitude. Low-cadence cycling is a deep aerobic stimulus that recruits some fast fibers; it is not a strength stimulus, and it does not replace the gym. If durability and finishing power are the goal, heavy gym strength training has the better evidence.
When It Does Seem to Help
There is a real exception worth stating. Hebisz and Hebisz (2024) trained 24 well-trained women on a polarized plan and found the group doing their hard intervals at 50-70 rpm improved VO2max more than the freely chosen cadence group, and only the low-cadence group significantly raised maximal aerobic power. This matches the review's one caveat: any benefit shows up at moderate-to-maximal intensity, not in easy grinding. Low cadence may be a useful way to add a strength-flavored quality to your hard intervals -- not a base-mile technique.
Example
Example: The Force Numbers That Settle the Gym Question
The clearest way to see why low-cadence riding cannot replace the gym is to put the joint forces side by side.
On the bike. Ericson and Nisell measured peak patellofemoral compressive force during cycling at 905 N -- about 1.3 times body weight -- at 120 W and 60 rpm. Push the power higher and it rises, but it stays in the low single-digit multiples of body weight.
In the gym. During a maximal isokinetic knee extension, the same research group measured patellofemoral force at roughly 12 times body weight.
| Movement | Peak knee compressive force |
|---|---|
| Cycling, 120 W, 60 rpm | ~1.3 x body weight |
| Maximal isokinetic knee extension | ~12 x body weight |
An order of magnitude separates them. Whatever the pedal stroke is doing to your fast-twitch fibers, it is not delivering the mechanical tension that drives a strength adaptation. So the "on-bike strength" claim collapses on the physics: you can grind all day and never approach the load a squat rack applies in one rep.
The flip side is the risk. That same force, applied thousands of times per session at a deep knee-flexion angle, is exactly what irritates a patellofemoral joint. Bini and Hume found the compressive force was 29% higher at 70 rpm than at 90 rpm for the same effort. So low-cadence work sits in an awkward spot: too little force to build strength, but enough repeated force to threaten a vulnerable knee.
Practical Rules
Practical Rules
-
Do not use it to replace the gym. The forces are roughly a tenth of what a strength movement produces, and heavy strength training has the stronger evidence for the durability and finishing-power benefits people hope to get from grinding. If you want a strength stimulus, lift.
-
Do not expect it to beat normal cadence at easy or moderate intensity. The cleanest trial found 40-rpm intervals produced nothing while free cadence raised VO2max. If the intensity is moderate, spinning is the better use of the session.
-
If you use it, pair it with high intensity. The one setting where low cadence outperformed free cadence was hard intervals in a polarized plan. Reserve big-gear work for your quality sessions, not your base miles.
-
Treat it primarily as specificity. If you race at low cadence -- long climbs, time trials in a big gear, mountain biking -- practising at that cadence trains the exact neuromuscular pattern you compete in. That is a legitimate reason to do it, separate from any metabolic claim.
-
Respect the knee, and screen yourself out if needed. Force at the patellofemoral joint is 29% higher at 70 than at 90 rpm for the same power. Anyone with a history of anterior knee pain, patellar tendinopathy or chondromalacia should not do high-torque low-cadence work without a proper bike fit and, ideally, clearance. This is not a nice-to-have caution -- the damage is cumulative.
-
Warm up thoroughly and keep the power sane. Knee force scales with workload, so a low cadence plus very high power is the worst combination. Build into it, keep it sub-threshold unless a study-style protocol calls for more, and stop at the first sign of joint pain -- muscle burn is the target, joint pain is not.
-
Stop it well before a race. High-torque work depresses your ability to produce power at racing cadence through peripheral fatigue; leave two to three weeks between the last real low-cadence block and a key event.
Evidence Base
Evidence Base
The thesis rests on a systematic review, which is the right kind of source. Hansen and Ronnestad (2017) is not a single study with a favorable result but a synthesis of the controlled trials, and its conclusion -- no strong evidence for a benefit -- is the most defensible position available. Its own caveat, that any benefit is confined to moderate-to-maximal intensity, is preserved in this article rather than smoothed over.
The mechanistic support is real but narrow. Ahlquist (1992) is a clean biopsy study, but it is eight men and it measures glycogen depletion, not performance. It proves fast fibers are recruited more; it proves nothing about whether that makes you faster. Treating fiber recruitment as if it were a performance outcome is the central overreach in the popular version of this topic.
The strongest "pro" study is confounded. Paton (2009) is cited everywhere as evidence for low cadence, and its own abstract concludes low cadence was "probably more effective." But the protocol alternated cycling with explosive single-leg jumps in both the intervals -- a potent neuromuscular stimulus that has nothing to do with cadence. The result may be real; the attribution to cadence is not supported.
The counter-evidence is clean. Kristoffersen (2014) matched work and isolated cadence, and the low-cadence group gained nothing while free cadence gained VO2max. That is the study design the others lack, and it points the opposite way to the folklore.
The knee data are consistent and specific. Bini and Hume (2013) and Ericson and Nisell (1987) both quantify patellofemoral force, and they agree: lower cadence and higher workload raise it, and the on-bike peak (about 1.3 body weights) is far below the gym (about 12). Those are small samples -- 12 and 6 participants -- but the physics is not controversial and the direction is unambiguous.
One study genuinely supports the method, with limits. Hebisz and Hebisz (2024) is a real positive result, but it is 24 women, one polarized protocol, and it works specifically because the low cadence was applied to high-intensity intervals. It supports "low cadence can help your hard sessions," not "grind your base miles."
Two limitations run across the literature. Samples are small throughout -- 6, 8, 12, 18, 22, 24 participants -- so specific percentages should be read as indicative. And the sex and level of the participants vary enough that findings do not transfer cleanly between, say, veteran men and well-trained women; the one clearly positive trial was in women, and the clearly negative one in older men, which may not be a coincidence.