What Cadence Should You Ride At?

Cadence is the most over-prescribed number in cycling. "Ride at 90 rpm" is repeated so often that it reads as settled science, and the research does not support it as a rule. What the research does support is more interesting, and more useful for someone training indoors where cadence is trivially easy to control.

What the published research actually found

The efficiency studies point consistently in one direction, and it is not the direction most cyclists expect. Summarising road.cc's review of the literature:

Suzuki (1979)measured gross efficiency in endurance athletes falling from 23.3% at 60 rpm to 19.6% at 100 rpm. Coast et al (1986)found the best cadences for trained racers at 85% of VO2 max were either 60 or 80 rpm. Leirdal and Ettema (2011)found efficiency highest around 10 rpm below the athletes' own freely chosen cadence. Nimmerichter et al (2015)tested 60 against 90 rpm on level and uphill terrain in trained cyclists and found 60 rpm more efficient in both conditions.

Field data supports the same conclusion: in one study of trained cyclists, gross efficiency was significantly higher at 60 rpm (20.6% ± 1.7%) than at 90 rpm (18.1% ± 1.3%). road.cc's summary of the whole body of work is blunt: most research shows about 60 rpm as the most efficient.

We have not run any of this testing ourselves, and we are not going to pretend otherwise — every figure above is attributed to the published source it came from.

So why does nobody ride at 60 rpm?

Because efficiency measures only one cost, and it is not the cost that limits a cyclist.

Force per pedal stroke.To produce the same power at 60 rpm as at 90 rpm, you have to push roughly half again as hard on each stroke. That load is borne by muscle and connective tissue, and it accumulates. A long effort at a low cadence leaves your legs wrecked in a way that the same effort at a higher cadence does not, even though it cost less oxygen.

Substrate use. road.cc notes reduced carbohydrate oxidation at higher cadences as one of the practical advantages, which matters on long rides where glycogen is a finite resource.

Responsiveness.A lower perceived force at the pedals makes it much easier to respond to an acceleration — a bridge, a surge, a steepening gradient. In a group or a race that is not a nicety, it is the whole point. At 60 rpm you are already near your force limit, so there is nothing left to add instantly.

It is uncomfortable.road.cc's own framing is that 60 rpm is not sustainable for most cyclists because there is more to riding a bike than being efficient. Riders self-select higher cadences for a reason, and Leirdal and Ettema's finding — efficiency peaks about 10 rpm below your freely chosen cadence — suggests the body is already making a sensible trade rather than getting it wrong.

What to look for in a cadence sensor

If you want to work with cadence at all, you need to be able to see it. On a smart trainer with measured power, cadence often comes free — but not always, and not accurately, because many trainers estimate it from flywheel behaviour rather than measuring crank rotation. A dedicated sensor measures the crank.

Both wireless protocols. Bluetooth and ANT+. Bluetooth is what Apple TV, iPad and Android need; ANT+ is what lets a PC or Mac read unlimited sensors through one dongle. ANT+ vs Bluetooth for Zwift explains why that matters, and it matters especially for a cadence sensor, because on Apple TV a cadence sensor is often the device that takes you past the two-Bluetooth-connection limit.

Magnet-less. Modern sensors read crank rotation with an accelerometer, so there is no magnet to align on the frame and nothing to be knocked out of position. Every sensor we recommend is magnet-less.

A coin cell with a published life. CatEye claims about 10 months at an hour a day; Garmin claims up to around 12 months. A sensor you replace a battery in annually is a sensor you never think about.

A water or dust rating. A crank-mounted sensor on an indoor trainer sits directly under a sweating rider. IPX7 is the figure to look for.

A mount that moves. A rubber band or strap around the crank arm transfers to another bike in seconds, which matters if you have a trainer bike and an outdoor bike.

The short answer

Cadence sensors worth buying

#	Product	Best for	Score	Price
1	CatEye CDC-30 The cadence sensor to buy if you just want a number you can trust without thinking about it again for most of a year.	Best cadence sensor overall	8.9★★★★☆8.9 out of 10	$35.95 · View on Amazon $39.95Save 10% #ad We may earn a commission · Price as of October 5, 2026
2	Garmin Cadence Sensor 2 Buy this if your head unit or watch is a Garmin and you want the sensor to simply appear and work.	Best for Garmin head units and watches	8.7★★★★☆8.7 out of 10	$67.99 · View on Amazon #ad We may earn a commission · Price as of October 5, 2026
3	Magene S3+ The cheap way to get a cadence number, with the useful quirk that it can be a speed sensor instead.	Best value, and switchable to speed	8.5★★★★☆8.5 out of 10	$16.99 · View on Amazon #ad We may earn a commission · Price as of October 5, 2026

What cadence to actually ride

Start from your freely chosen cadence. Ride normally for a few sessions and look at the average. That figure is the output of a long unconscious optimisation, and it is the right default. If it is somewhere in the 80s or low 90s, you are in company with most trained cyclists and there is nothing to fix.

Do not chase 90 rpm because you read it somewhere.Forcing a cadence 15 rpm above your natural one raises your heart rate for the same power and feels worse without producing anything. If you want to change your natural cadence, that is a slow adaptation, not a decision.

Vary it deliberately in training — this is the actual use.Indoors, cadence is the easiest variable in the room to control, and varying it changes what a given wattage asks of you. A sweet-spot interval at 95 rpm is an aerobic session. The same interval at 65 rpm is a muscular one. Low-cadence work at 50 to 60 rpm while holding the same power is a recognised way to load the muscle without raising the metabolic cost; high-cadence work above your natural figure trains the neuromuscular coordination that lets you spin smoothly when fatigued. Both are tools.

On climbs, cadence falls — do not fight it. On a steep gradient, gearing runs out before cadence preference does, and the research does not suggest you are losing anything by riding at 70 on a climb.

Watch cadence during intervals, not just power. A cadence that drifts down through a long interval while the wattage holds is a sign of muscular fatigue arriving. On a trainer in ERG mode this is especially worth watching, because the trainer will keep holding the wattage while your cadence collapses. ERG mode explained covers the spiral-of-death failure mode that follows.

Cadence, ERG mode and the spiral of death

This is the one cadence problem unique to indoor riding and worth knowing before it happens. In ERG mode the trainer holds a fixed wattage regardless of your gear, which means that if your cadence drops, the trainer increases resistance to maintain the target, which makes your cadence drop further. The result is a stall. The fix is to anticipate it: pick a gear that puts you at a comfortable cadence before the interval starts, and if you feel the cadence falling away, spin up rather than pushing harder.

That is also the single best argument for owning a cadence sensor rather than relying on your trainer's estimate: a real crank measurement gives you the early warning. Our cadence sensor vs power meter guide covers which to buy first, and cadence sensor for a stationary bike covers the case where the bike has no crank sensor at all.

Does cadence affect your power numbers?

It can affect what your power meter reports, slightly. Some measurement designs are more sensitive to very low or very high cadences than others, which is why comparative accuracy testing runs multiple cadences rather than one. It is a small effect and not a reason to choose a cadence, but it is a reason not to be surprised if a trainer and a power meter agree less well at 50 rpm than at 90. How accurate are power meters covers the whole picture.