How Accurate Is a GPS Speedometer? Real Numbers, Tests & Limits

1. The short answer: ±1–3 km/h when it matters most

With open sky, a settled fix, and steady movement, a modern phone GPS speedometer lands within about 1–3 km/h (0.6–1.9 mph / 0.5–1.6 knots) of true speed. That is not a best-case lab figure — it is what you get on a highway cruise, a steady bike ride, or a straight run once the accuracy badge has settled to ±15 m or better.

Three qualifiers decide whether you get that number today: sky view(buildings hurt more than clouds), steadiness (a 10-second average beats any 1-second spike), and warm-up (the first ~30 seconds are always the worst). If any of the three is missing, accuracy degrades predictably — which is exactly what the tables below quantify.

2. Real accuracy numbers by activity

These are field-typical ranges for a phone from roughly 2018 onward with high-accuracy location on, after a 30-second warm-up. “Typical error” means the 10-second average versus true speed — not the worst one-second spike, which is always larger.

Notice the pattern: faster and more open is easier, slower and more enclosed is harder.At 90 km/h you travel 25 metres per second — far above position noise — so speed is rock-solid. At a 5 km/h stroll you travel 1.4 metres per second, smaller than the noise itself, so the live number breathes even when the average is honest. Try the bike mode,car mode orboat mode in knots to see the right unit for each case.

3. What the ± meters badge actually tells you

Every fix carries an accuracy estimate in meters — the receiver's own uncertainty radius forposition, shown as a badge on our live speedometer. It is not speed error directly, but it is the fastest honest signal you have:

• ≤ 15 m — green, trustworthy. Settle here before judging anything. Highway-grade speed accuracy (±1–2 km/h) lives in this zone.
• 16–50 m — amber, use the average. Common downtown or under trees. The live digit will wander; the 10-second average is still useful. Move a few metres toward open sky if you can.
• > 50 m — wait or move. Likely indoors, battery-saver positioning, or a cold start. Do not quote the number — give it 30 seconds outdoors first.

One subtlety worth knowing: speed stays usable slightly longer than position.Errors shared between consecutive fixes — ionospheric delay is the classic example — shift both dots the same way and cancel out in velocity. Your map dot can sit metres off the road while speed stays within 1–2 km/h. That is why a restless dot with a steady average is normal, not broken.

4. GPS vs your dashboard: who is telling the truth?

Compare the two at a steady cruise and the dial almost always reads higher. Drivers assume the phone is wrong. Usually it is the other way round: regulations in the EU, US, India and elsewhere allow speedometers to read high by a margin (never low), and manufacturers use that margin as a safety buffer.

So on an open highway with a green badge, GPS is frequently closer to physical truth than the needle — which is why fleet loggers and cycling computers trust it. But “closer to truth” is not “certified”: the dashboard is type-approved, always available, and stable in tunnels, while GPS needs sky and carries no certificate. Use them as a pair. If both agree within a few km/h at cruise, both are healthy. If they disagree by more than ~10% repeatedly in good signal, suspect tyre size changes, a failing wheel sensor, or an overheating phone — and investigate. The full comparison with test protocol is in our car speedometer guide, and the legal fine print in the disclaimer: the dial is the only legal instrument, GPS is never a defence in a speeding dispute.

5. The 7 things that actually move the needle

Ordered by damage, from total failure to mild noise:

1. 1. No sky, no fix. Tunnels, underground car parks, deep indoors — the receiver has nothing to time. An honest tool shows Searching instead of guessing. Ours does.
2. 2. Multipath — the city killer. Signals bounce off glass towers and cliffs and arrive late, wandering fixes 20–40 m. This is the cause of nearly every “it jumped to 68 in a 50 zone” story downtown.
3. 3. Body, case and windscreen blocking. Phone in a pocket or bag, thick metal cases, and heat-reflective windscreens cost 10–20 m. Mount on the dash or handlebars, screen up, with a sky view.
4. 4. Cold start. The first ~30 seconds after Start are spent downloading orbital data. Errors of ±15–40 m are normal here — wait for the badge to settle, then Reset so averages start clean.
5. 5. Power saving. Battery savers silently downgrade to Wi-Fi/cell positioning (±500 m+), which cannot produce meaningful speed. Grant Precise Location and exempt your browser on long rides.
6. 6. Slow speed itself. Below ~8 km/h the motion per second drops under the noise floor. The average stays honest; the live digit breathes. Walk tests should use distance over a measured kilometre, not second-by-second speed.
7. 7. Space weather, slightly. Ionospheric delay adds ±2–5 m, mostly modelled away — and dual-frequency phones (L1+L5, common since ~2018) measure and subtract most of it directly. Real, but rarely the limiting factor.

And what barely matters: clouds, light rain and cold. GPS radio at 1.5 GHz sails through weather with centimetres of extra delay. A grey rainy day under open sky still delivers ±5–10 m. When someone blames clouds downtown, the buildings did it.

6. Where GPS speed struggles — and what to do instead

• Street canyons: judge the 10-second average, ignore spikes, step toward intersections and open stretches. Two phones side by side can honestly disagree by a few km/h here — both are behaving correctly for their hardware.
• Forests and valleys: canopy and cliffs block half the sky. Speed on straight fire roads stays decent; twisty single-track distance will read slightly short as corners get cut between 1-second fixes.
• Trains: metal carriages attenuate signals; sit by a window, wait longer for lock, and expect ±2–5 km/h at speed. Tunnels drop the fix entirely — watch ours pause honestly instead of freezing on a stale number. See the train speed test guide.
• Stop-start traffic: constant acceleration plus blocked sky at junctions is the hardest combination. Average and max over the trip stay meaningful; second-by-second speed at the lights does not.

7. How this site measures speed (and why it matters for accuracy)

Not every GPS page measures the same way. Ours uses a strict order of preference, documented inhow GPS speed works:

1. Doppler velocity first. When the browser exposes coords.speed, we display it directly — a 3D velocity vector computed from carrier-frequency shift, independent of position subtraction. This is why our highway number stays steady while the map dot wobbles.
2. Haversine fallback with smoothing. When browsers withhold speed, we compute great-circle distance between consecutive fixes divided by elapsed time, smoothed over the last few fixes. Genuine acceleration shows through in a second or two; single-fix reflections get averaged down.
3. No fake precision. No fix means Searching — never a coasted guess. Coordinates never leave your device; read the privacy policy.

Paid apps share the same OS location feed and the same physics — money buys logs, maps and exports, not a better Doppler shift. Our free page reads the identical sensor.

8. Test your own phone in 5 minutes

One experiment beats ten forum threads. As a passenger — never handle the phone while driving — pick a straight stretch with open sky:

1. 1. Warm up. Open the live speedometer, tap Start, and wait until the badge reads ±15 m or better (usually 20–30 seconds).
2. 2. Reset. Hit Reset so average and max start clean from the settled fix.
3. 3. Hold steady 60 seconds. Ask the driver to hold cruise at 80 km/h. Note the 10-second GPS average, ignoring spikes. Typical result: dial 80, GPS 76–78.
4. 4. Repeat once. Do a second speed (e.g. 60 or 100 km/h) to confirm the offset is proportional — that confirms a healthy dial margin rather than a GPS fault.

Cyclists: ride one measured kilometre steadily and compare the distance tile. Runners: compare a 400-metre track lap. Same protocol — warm up, reset, hold steady, read the average — because the error physics is identical and only the scenery changes. Convert unfamiliar units withkm/h → mph ormph → km/h.

9. Five habits for honest numbers

1. Seek open sky first. A footpath beats a balcony; a park beats a downtown canyon. Buildings are error source number one.
2. Mount it, don't swing it. A swinging hand adds phantom metres per second. Dashboard cradle, handlebar mount, or flat and still at chest height.
3. Wait out the warm-up, then reset. 20–30 seconds after Start; reset once the badge turns green so session stats start clean.
4. Read averages, not spikes. Ten seconds of steady average is the truth; a one-second peak is usually a reflection.
5. Keep high-accuracy location on. Precise Location granted, battery-saver exemptions for long sessions, phone shaded from direct-sun overheating which throttles updates.

Bottom line for 15 Oct 2026 hardware: on open road at steady speed, GPS speed is a±1–3 km/h instrument — often more truthful than the optimistic dial, always subordinate to it legally. Learn your car's personal optimism number once with the test above, and you will read both instruments correctly for years. For the deeper physics behind these numbers, read how a GPS speedometer works.

FAQ

Questions this article answers

Try the accuracy yourself

The fastest way to feel everything above is a two-minute outdoor test: start the live tool, watch the badge settle to green, and compare the 10-second average against a steady pace.