Frequently asked questions

We compile every published specification that matters for the decision — resistance ceiling, simulated gradient, flywheel mass, claimed power accuracy, connectivity and noise — read the manufacturer manuals, aggregate disclosed owner feedback where it is properly cited, then do the cost-per-watt and accuracy-per-dollar math and apply the same rubric to every model. The pick that wins on the criteria we published is the pick we recommend. Because the inputs are all on the page, you can re-run the comparison and weigh it differently if your priorities differ.

No, and we say so plainly. The number of units we claim to have bench-tested is zero. Everyone in this category says they tested twenty trainers in a lab; we have not tested any. What we did instead is read every manual, compile every published spec and accuracy figure, and show the math so you can check it. We think an honest 'here is what we compiled' beats an unverifiable 'trust us, we tested it.'

Because we only ever show a live price. Our pricing comes from a real-time layer rather than from memory, and when a current price is not available for a product we show 'Check price' and send you to the retailer for the real number. An out-of-date price is worse than no price, so we refuse to print a stale one.

Affiliate commissions. When you follow one of our outbound links to Amazon or another retailer and buy something, we may earn a commission at no extra cost to you. That is the entire business model, and it is what keeps the site free to read.

No. Picks are determined by the published rubric — specs, math and value — applied identically to every model. A product that pays a higher commission gets no advantage, a product that pays us nothing is eligible to win, and we write the 'skip this' verdict whenever it is the honest one. We accept no sponsored placements and no free units.

The price we display is pulled live at the time of the request, but prices change constantly and the retailer's own page is always the source of truth. Treat our number as a strong signal, then confirm the final price, availability and terms on the retailer's page before you buy.

A wheel-on trainer presses a roller against your rear tire; it is cheaper and quicker to set up, but it is noisier, less accurate, prone to tire slip under high power, and wears your tire. A direct-drive trainer removes the rear wheel and drives the cassette directly, which is quieter, far more accurate, handles big power without slip and gives better road feel. If the budget allows, direct drive is the upgrade most riders feel immediately.

Not strictly — Zwift only needs a way to broadcast your power and to control your ride. You can run Zwift with a 'dumb' trainer plus a speed sensor (using its power curve) or a separate power meter. But a smart trainer is the setup Zwift is built around: it reports accurate power and automatically adjusts resistance for gradients and ERG-mode workouts, which is a big part of why indoor riding feels worth it. See our guide to the best smart trainers for specifics.

For most riders, a claimed accuracy of plus or minus 2 to 2.5 percent is plenty — the difference between that and plus or minus 1 percent is often within the noise of your own day-to-day variation. Accuracy matters most if you race, chase precise FTP numbers, or want your indoor and outdoor power to line up closely. For general training and Zwift, do not overpay for the last half-percent; put the money toward road feel instead.

For indoor training, no — the smart trainer already measures your power. A separate power meter earns its keep when you also ride outdoors and want consistent power data on the road, or when you want a second source to cross-check and calibrate your trainer. If you never ride outside, the trainer alone is enough.

A good fan. Nothing improves indoor rides more per dollar — without airflow you overheat, your heart rate drifts up, and your power drops, so a strong fan effectively buys you watts. After that, a thick trainer mat (to protect the floor and cut vibration) and a front-wheel riser block if you are on a wheel-on setup are the next best small purchases.

Flywheel mass drives road feel — the sensation of inertia as you pedal and coast. A heavier flywheel (or a lighter one paired with electronic simulation) carries momentum more like riding outdoors, so the pedal stroke feels smooth rather than sticky, especially when you stop pedaling or surge. It does not change your power numbers, but it is a major reason premium trainers feel better to ride.

Simulated gradient is the steepest climb a smart trainer can physically replicate by adding resistance — for example 'up to 20 percent.' Higher ceilings make steep Zwift and route climbs feel realistic. For most riders anything from about 12 percent up is plenty, since you rarely hit the trainer's ceiling; chase a very high number only if you specifically want the steepest virtual climbs to bite.

Ideally both, and most current smart trainers include both. Bluetooth connects easily to phones, tablets and computers and is what most people use with Zwift on a tablet. ANT+ is a robust older standard favored by dedicated head units and lets many devices read the same signal at once. If a trainer only had one, Bluetooth is the more broadly useful today — but you rarely have to choose.

Flagship roundups and comparisons get a quarterly refresh — re-checked prices and availability plus any newly shipped models — and we update on events, revising affected pages within days when a significant product launches or is discontinued. Every page shows a visible publish and last-updated date so you always know how fresh it is.

Stephen V., an enthusiast who is genuinely into indoor training — no lab coat, no tenure claims and no invented credentials. The whole method is one person reading the manuals, compiling the published specs and doing the watt math in the open. That honesty about who we are is the point, not a disclaimer.

Email us. Corrections are one of the best sources of accuracy we have, so if a spec looks wrong, a price is off, or a pick seems misjudged, we want to hear it. Every substantive correction is logged and the page is re-dated when we fix it.