How to choose an electric bike

Decision 1 — What is the bike actually for?

Almost every bad e-bike purchase starts the same way: with a specification sheet rather than with a sentence. Before you compare anything, finish this sentence honestly — "I am buying this bike so that I can ___."

The answer eliminates more of the market than any filter on a retail site. A bike bought to replace a four-mile car trip to the shops has almost nothing in common with a bike bought to ride forest tracks at the weekend, and the two are usually sold on the same page with the same photographs.

The five common jobs

• Commuting. Repeatability is everything. You need range with margin, lights, mudguards, a rack, and a bike that starts every morning without attention. Top speed matters less than the ability to do the same trip 200 times.
• Trail and loose ground. Traction, suspension, braking and ground clearance move to the front. Range matters less because the rides are shorter and slower.
• Cargo and errands. Payload, frame stiffness, a low centre of gravity and brake capacity dominate. Speed is close to irrelevant.
• Leisure riding. Comfort, upright position and tire volume matter more than anything electrical.
• Replacing specific car trips. The strictest brief of the five, because the bike has to beat the car on convenience rather than only on cost. Parking, locking and load capacity decide it.

Write your sentence down. Every later decision in this guide refers back to it, and any specification that does not serve it is noise.

The order that separates these twenty-two bikes

When a shortlist is three bikes from one brand at one price point, watt-hours come first, because everything else on the sheet is the same. Twenty-two bikes across four brands from $639.99 to $2,999 is a different problem. The specifications that actually separate them, in the order worth deciding, are: sensor type, torque, brakes, certification, watt-hours.

The reason that order works is that the first four are separable on most of the field. Sixteen of the twenty-two publish a torque figure and thirteen publish a sensor type. Velotric publishes both for all six of its bikes. VTUVIA publishes a sensor type for five of six and a torque figure for five of six, though not the same five. EUNORAU publishes a torque figure for all five of its bikes but a sensor type for only two. AMYET publishes neither, for any of its five models. That means a large part of the field can be sorted before you look at a battery at all — and it means watt-hours, which used to be the first question, becomes the last one.

Below about $1,200 you are buying watt-hours. Above it you stop buying watt-hours and start buying brakes, sensors, gearing, suspension, certification and — most of all — published specifications.

Decision 2 — Sensor type

This is first because it changes how the bike feels on every single ride, and because it costs nothing to check.

A cadence sensor watches whether the pedals are turning. When they turn, the motor applies the assist level you selected; push harder and nothing changes. A torque sensor measures how hard you are pushing and scales the motor to match, so assist rises when you stand on a climb and falls when you soft-pedal. Riders generally describe the second as feeling like a bicycle and the first as feeling like a throttle with pedals attached.

Thirteen of the twenty-two bikes here state a sensor type. VTUVIA states a cadence sensor on the SF20H, a V-Adapt torque sensor on the Zeal LT7, a switchable torque-or-cadence system on the SX20, a torque sensor on the SN100, and a torque sensor on the Gemini — named inside the colour option values rather than in a specification line, which is an odd place to put it but is where VTUVIA puts it. EUNORAU states a torque sensor on the META26 and on the SPECTER-S. Velotric states one on all six of its bikes, and inverts the usual order at the bottom of its range: the torque sensor is on the cheapest bike, the $1,399 T1 ST Plus, while the dearer Nomad 1 Plus and Packer 1 both use a cadence sensor. The three Discover models sit between the two positions with a combined torque-and-cadence system, switchable on the Discover 2 as published.

The practical rule: if you will pedal, a torque sensor is the single upgrade most likely to make you ride the bike more. If you plan mostly to use the throttle, a cadence sensor costs you nothing.

Decision 3 — Torque, and the terrain question behind it

Gradient decides more of the specification than any other input, and torque in newton-metres is the number that predicts it. Watts describe the rate of doing work; newton-metres describe the shove. Climbing is a low-speed, high-force problem, so it is the shove that matters.

The ladder runs: 40 Nm on the Velotric T1 ST Plus, 55 Nm on the EUNORAU META26, 65 Nm on the VTUVIA Zeal LT7, 75 Nm on the Velotric Nomad 1 Plus, Packer 1, Discover 2 and Discover 3, 85 Nm on the VTUVIA SF20H and SX20, 92 Nm on the EUNORAU FLASH Lite in its rear-hub configuration, 100 Nm on the VTUVIA Gemini, 160 Nm from the Bafang G510 mid-drives in the EUNORAU SPECTER-S, the EUNORAU SPECTER-ST and the VTUVIA FMB, and 220 Nm from the Truckrun mid-drive in the FLASH. Two figures sit outside that single line: Velotric's Discover M states 100 Nm from a mid-drive rather than a hub, and the FLASH Lite states 184 Nm in its dual-hub AWD configuration, which is two motors added together rather than one. Mid-drive figures are quoted at the crank and pass through the gears, so they are not directly comparable with hub figures at the wheel. There are five mid-drives here now, and they are no longer all from one brand: EUNORAU's FLASH, SPECTER-S and SPECTER-ST, VTUVIA's FMB and Velotric's Discover M, with the FLASH Lite offering a mid-drive as a motor-position option.

The top of the ladder outside the mid-drives has moved. It is now the Gemini's 100 Nm — from a motor VTUVIA never places — with EUNORAU's FLASH Lite rear hub at 92 Nm the most any bike here states from a confirmed single hub, ahead of the 85 Nm of the SF20H and SX20.

If your terrain is flat, most of this is safely ignorable and the money is better spent on comfort. If it is not, our hills guide and the motor wattage guide take the mechanism apart properly.

Decision 4 — Brakes

Brakes are where a specification sheet's silence is loudest. Hydraulic disc brakes self-adjust as the pads wear, need far less lever force, and matter more the heavier and faster the bike is. Mechanical discs are cheaper and serviceable with a cable and a hex key.

The rule is simple: where a listing does not say hydraulic, read it as mechanical. Brands that fit hydraulics say so, because they cost more.

Eighteen of the twenty-two carry a hydraulic claim. Every VTUVIA bar the FMB states hydraulic discs, as does every EUNORAU and every Velotric, and many of them state a rotor size — 180mm on the SF20H, Zeal LT7, SX20, META26, FLASH, SPECTER-S and Velotric's Packer 1, Discover 2, Discover 3 and Discover M, and 160mm on the Velotric T1 ST Plus. Velotric states hydraulic discs on the Nomad 1 Plus without naming a rotor size, and VTUVIA does the same on the Gemini and the SN100. VTUVIA goes further on the SF20H and names the caliper brand and the rotor material; Velotric goes further again on the Packer 1, which is the only bike here with four-piston calipers, and names TEKTRO on all three Discovers. On the AMYET side the S8 and the K10 carry hydraulic claims; the EB26 is described as mechanical disc, and the V9-G60 and the Ares are described only as "dual disc" and "disc brakes", which we read as mechanical for exactly the reason above. VTUVIA publishes no brake specification at all for the FMB, which on a $2,999 full-suspension bike is a conspicuous silence. Two of the bikes here — the META26 and the SPECTER-S — also put motor cut-off switches in the brake levers, as do all three Discovers, so the motor stops the instant you reach for a lever. On a throttle bike that is worth having.

Decision 5 — Certification

Two UL standards appear on e-bike listings and they are not interchangeable. UL 2271 applies to the battery pack alone. UL 2849 is broader: pack, charger, controller and motor evaluated together as one system. Most serious lithium failures come from a mismatch somewhere in that chain rather than a cell failing in isolation, which is why the whole-system standard is the stronger claim.

This has left the realm of theory. New York City's Local Law 39 is the best-known statutory example, and beyond specific legislation a growing number of landlords, building managers, employers and insurers ask for UL 2849 before permitting indoor storage or charging.

Across the twenty-two bikes, the paperwork is uneven, and seven state both UL numbers. All six Velotrics state both UL 2271 and UL 2849, and add UL 2580 for the cells and ISO 4210 for the bike — the most complete paperwork on this site, and still the only brand here where every model states both UL numbers, the three Discovers included. The VTUVIA SF20H states both as well, which makes it the only bike outside the Velotric range to do so; at $1,399 it ties the T1 ST Plus as the cheapest bike here stating both. AMYET states UL 2849 certification for the V9-G60 and UL 2849 compliance for the EB26. EUNORAU states UL certification for the SPECTER-S without naming which standard, and VTUVIA does the same for the SN100 battery.

Decision 6 — Watt-hours, and how many you actually need

Only now does the battery matter, and it matters as a threshold rather than as a ranking. Ignore the headline range on the listing entirely and work from stored energy: multiply battery voltage by amp-hours to get watt-hours, the one capacity figure that compares fairly across brands and voltages.

Take your longest regular ride, multiply by 25 watt-hours per mile for realistic mixed riding on a heavy fat-tire bike, then add 20% as reserve. This is an estimate from published capacity, not a tested result.

Worked through: a 14-mile round trip to work needs 14 × 25 = 350Wh, plus 20% for cold mornings, headwinds and a pack that will not be new forever, so roughly 420Wh. One bike here does not clear that: the Velotric T1 ST Plus carries 36V 9.8Ah, which is 352.8Wh, the smallest pack on this site by a wide margin. It also runs the narrowest tires we cover, so it draws less per mile than the 25Wh the sum assumes — but it has the least in reserve of anything here. Every other bike with a published capacity clears 420Wh with room, the next smallest being the Zeal LT7's 48V 13Ah, which is 624Wh. Now make it 32 miles: 32 × 25 × 1.2 = 960Wh, and suddenly only the AMYET V9-G60 at 960Wh, the AMYET Ares 2 at 1,040Wh, the AMYET S8 at 1,200Wh and the multi-battery EUNORAU bikes are in the conversation at all.

That is the whole point of doing the arithmetic first. It converts an open field into a short one. Our battery and range guide sets out where the 25Wh figure comes from and works through all twenty-two packs, including the bikes that take extra batteries.

Decision 7 — Where will it live?

This is the decision that produces the most regret, and it is almost never on a comparison table. Go and measure things before you buy: the doorway, the hallway, the lift, the distance from the front door to where the bike will actually sit, the boot of your car, and whether your rack is rated for the weight, because many boot-mounted racks are not.

Weight is published often enough to sort by, though less often than it used to be — eleven of the twenty-two bikes here give a figure. Velotric lists the T1 ST Plus at 39 lb, the lightest bike on this site by a wide margin; VTUVIA lists the SF20H at 60 lb, the Zeal LT7 at 62 lb and the SX20 at 70 lb; Velotric lists the Discover 3 at 61 lb; EUNORAU lists the META26 at 68.4 lb; Velotric lists the Nomad 1 Plus at 70 lb for the step-thru and 71 lb for the high-step, and the Packer 1 at 75 lb, which is the heaviest hub-motor bike here without being anywhere near the heaviest bike; and EUNORAU lists both the FLASH and the SPECTER-S at 50 kg, which is approximately 110 lb.

At 110 lb the FLASH and the SPECTER-S are not bikes you lift; they are bikes you keep at ground level and ride from the door. Wheel size feeds into the same decision — a 20-inch bike fits spaces a 26-inch or 27.5-inch bike will not, which our 20 vs 26 inch guide covers. And one storage question is really a safety question: if the bike will charge in the interior hallway of a shared building, Decision 5 may have already made your shortlist for you.

Decision 8 — What are the rules where you ride?

Around forty US states have adopted a three-class framework.

• Class 1. Pedal assist only, no throttle, assist ends at 20 MPH. The widest path access.
• Class 2. Throttle permitted, still capped at 20 MPH.
• Class 3. Pedal assist to 28 MPH, usually no throttle above 20, and frequently barred from shared-use paths and some bike lanes.

Most states also apply a nominal motor cap, commonly 750W, for a machine to count as a bicycle at all. A bike assisting past 28 MPH sits outside all three classes in those jurisdictions, which can mean it is legally a moped, with registration, licensing and insurance implications attached.

Sorted by claim: the VTUVIA SF20H, Zeal LT7 and SX20 and the EUNORAU META26 and FLASH are stated at 20 MPH, which is Class 2 territory, and all six Velotrics default to 20 MPH as well. Velotric is unusually specific about what sits behind that number — the Packer 1 is stated as Class 2 and unlockable to 25 MPH, the T1 ST Plus as Class 1/3 with a limit adjustable between 12 and 28 MPH, the Discover 2 and Discover 3 as Class 1/2/3 with the same 12 to 28 MPH adjustment, and for the Nomad 1 Plus no class is stated at all. The Discover M states the class without stating a speed. The AMYET EB26 and K10 are stated at 28 MPH and the EUNORAU SPECTER-S at 28 MPH, all exactly at the Class 3 ceiling. The AMYET V9-G60, S8 and Ares are stated at up to 32 MPH, which is outside the framework in states that use it. VTUVIA notes that the SF20H ships factory-limited to 20 MPH and is capable of up to 28 MPH at full power — raising a limiter changes the class the bike rides under.

Find your state's e-bike statute, then check your local park or trail authority separately, because path rules are frequently stricter than road rules.

Decision 9 — Total cost, not sticker price

A $639.99 e-bike does not cost $639.99. It costs $639.99 plus consumables plus a battery plus whatever servicing you cannot do yourself, over however many years you keep it.

• A replacement pack. The battery is the only component guaranteed to wear out, typically after 500 to 800 charge cycles. Across the US market a replacement 48V pack of this size generally runs into the low-to-mid hundreds of dollars. Ask the price, and ask how long the brand commits to stocking it.
• Extra batteries, if the bike takes them. Five bikes here take them and all five are EUNORAUs: the META26's second pack, the SPECTER-S's second pack, the SPECTER-ST's second pack, and two additional packs each on the FLASH and the FLASH Lite. All are sold separately and all are a substantial cost on top of the sticker.
• Tires, pads and rotors. Four-inch fat tires are not cheap, and mass and speed both increase pad consumption.
• Assembly and first service. AMYET lists the G60 and EB26 as shipping 90% assembled and recommends professional installation for the S8; the K10 and Ares listings say nothing about assembly at all, as is true of every one of the ten newest listings here. A shop build is money well spent on anything carrying this much mass.

Warranty terms differ: AMYET states one year, VTUVIA a two-year frame warranty, EUNORAU two years. The better question is spares. A bike whose battery, display and brake parts you can still order in year three is worth more than an extra six months of cover with no parts channel. How we weigh unpublished specifications is set out on our how we review page.

How to read a spec sheet sceptically

Rated versus peak watts. When a brand leads with a peak figure it is describing the best few seconds of the motor's life. Compare rated against rated, always.

Any range figure beginning "up to". Maximum range claims are produced at the lowest assist level, on flat ground, with a light rider, in mild weather. Where a brand publishes two figures it is being more useful: AMYET states around 30 miles on throttle against up to 60 with assist for the EB26, and gives a 35 to 60 mile band rather than a single ceiling for the K10; VTUVIA publishes a separate throttle-only figure for the SF20H, SX20 and Zeal LT7, though for none of its three newest bikes; and Velotric does the same for five of its six, adding throttle figures of 60, 65 and 70 miles for the Discover 2, Discover 3 and Discover M. Velotric goes one better still and footnotes every range figure with the conditions behind it — a 187-lb rider, on PAS 1, on flat roads — which no other brand here does at all. VTUVIA's FMB claims a modest 30 to 50 miles, which is the least ambitious range claim on this site and reads the more honestly for it. Do the watt-hour arithmetic yourself anyway.

Compare-at pricing that never expires. A permanent strike-through price is a marketing device. Treat the current price as the price. Watch for the reverse, too: the FMB's compare-at price is *lower* than its selling price, which is a merchant data error rather than a discount, and we show no saving on it.

What is missing. Absences are information, and there are more of them than there used to be. AMYET publishes no torque, sensor type or bike weight for any of its five models. EUNORAU publishes no payload, sensor type, charge time or rider height for the FLASH or the FLASH Lite, and none of those bar charge time for the SPECTER-ST. VTUVIA publishes fork travel for the Zeal LT7 but not for the SX20, and publishes no payload, weight, top speed or sensor type for the FMB. Velotric publishes no charge time for any of its six bikes, no rider height for the Packer 1, and no weight for the Discover 2. Seven bikes here publish no payload figure at all, where among the original twelve only the FLASH did not.

What to ignore. Frame colour options, the number of assist levels, USB ports and "military-grade" anything. None of these decides whether you ride the bike in year two.

The buyer checklist

Keep this open in a tab while you compare. If you cannot answer a line from the listing, that is your question for the brand's support team.

1. The sentence. What is this bike for, and does the specification serve that job?
2. Sensor type. Torque, cadence, switchable — or unpublished?
3. Torque in newton-metres. Published, and how does it compare within its motor type?
4. Brakes. Does the listing say hydraulic, and is a rotor diameter given?
5. Certification. UL 2849 on the complete bike, UL 2271 on the pack, or nothing stated?
6. Watt-hours. Volts × amp-hours. At least your longest regular ride × 25, plus 20%?
7. Bare bike weight. Published, a shipping weight, or absent?
8. Payload rating, remembering it covers rider plus clothing plus rack plus cargo. Fifteen of these twenty-two publish one, running from 286 lb on the META26 to 440 lb on five Velotrics — the Nomad 1 Plus, the Packer 1 and all three Discovers. Seven publish no figure at all.
9. Legal class where you ride, including path rules.
10. Battery removable, and sold separately as a spare part.
11. Assembly. What arrives unassembled, and will you pay a shop?
12. Storage. Have you measured the space, the doorway and the car?

To see the framework applied to specific bikes rather than in the abstract, our best electric bikes ranking works through all twenty-two against exactly these criteria, and AMYET vs VTUVIA is the clearest illustration of what the published-specification gap costs. If you are shopping close to a payload ceiling, read the guide for heavier riders before you commit.