How GPS Location Accuracy Works in Web Browsers & Smartphones

When a web page calls navigator.geolocation.getCurrentPosition(), the browser asks the operating system for a position, and the system picks from several sources:

1. GNSS satellites (GPS, Galileo, GLONASS, BeiDou): need a view of at least four satellites. Outdoors with an open sky, GPS.gov puts typical smartphone accuracy at about 4.9 m (16 ft); near buildings, bridges and trees it gets worse.
2. Wi-Fi positioning: the device scans nearby Wi-Fi access points and compares them with a location database maintained by the platform provider. Accuracy is typically in the tens of meters and works indoors.
3. Cell towers: the device estimates position from the towers it can hear. Accuracy ranges from a few hundred meters in cities to several kilometers in rural areas.
4. IP geolocation: used when nothing better is available, for example on a desktop connected by Ethernet. The address is looked up in a database of internet provider allocations and is often off by many miles, sometimes pointing to a provider's regional hub in another city.

Phones usually blend these sources, so the same request can return a Wi-Fi estimate first and a sharper satellite fix a few seconds later.

The Geolocation API returns an accuracy value in meters. The current W3C specification defines it at the 68% confidence level (one standard deviation): an accuracy of 8 meters means there is roughly a two-in-three chance your true position lies within 8 meters of the reported point, and a real chance it lies somewhat farther out. Earlier drafts, and many references that copied them, described it as a 95% radius; the specification has since been changed to 68%. If the errors are roughly normal and the same in every direction, a circle about 1.6 times the reported value holds your true position about 95% of the time; doubling the value is a simple, slightly conservative bound.

A few practical consequences:

• Smaller is better. An accuracy of 5 meters suggests a satellite fix; 30 to 100 meters usually means Wi-Fi; several thousand meters points to cell towers or IP.
• The point is not the truth. The reported coordinate is the center of the estimate. Your real position can be anywhere inside the circle.
• Accuracy is self-reported. It is the device's own estimate. In difficult environments, such as between tall buildings, the true error can exceed it, and with a one-standard-deviation radius that happens on about one fix in three even in normal conditions.

The latitude and longitude finder shows the reported accuracy alongside your coordinates, so you can judge whether the fix is good enough before you copy it. If you are picking a point on a map rather than using your own position, see how to find coordinates on Google Maps; a map click has no satellite error, only the precision of where you click.

Even with a good receiver, several conditions degrade a satellite fix:

• Buildings and terrain. Tall buildings block satellites and reflect signals. Reflected signals travel a longer path, which biases the position. This is called multipath.
• Indoors. Roofs and walls weaken satellite signals, so indoor fixes often come from Wi-Fi instead.
• Tree canopy. Dense, wet foliage attenuates signals.
• Satellite geometry. When the visible satellites are bunched in one part of the sky, small timing errors produce larger position errors. Receivers express this as dilution of precision (DOP); a low DOP means good geometry.
• Cold starts. A receiver that has been off for a while needs time to identify satellites. Assisted GPS shortens this by downloading orbit data over the network.
• Power saving and permissions. Battery savers and approximate-location settings can stop the phone from using satellites at all.

Vertical accuracy is generally worse than horizontal accuracy, because all visible satellites are above you. The satellite signals themselves are very good: the U.S. government commits to a global average user range error of 2.0 m (6.6 ft) or less with 95% probability; most of the error you see comes from the receiver's surroundings. Browsers may return altitude as null; for ground height it is usually better to look up a terrain model with What Is My Elevation.

If an accuracy reading looks too large, try these steps in order:

1. Check permissions. Make sure the browser has location permission and, on iOS and Android, that precise location is turned on for the browser.
2. Turn on Wi-Fi. Even without connecting to a network, Wi-Fi scanning helps the phone locate itself quickly indoors.
3. Move to open sky. Step outside or near a window, away from tall walls.
4. Wait a few seconds and retry. The first answer is often a quick, coarse estimate. A second request after the receiver settles is usually sharper, so find your GPS coordinates again and compare the two accuracy values.
5. Check the accuracy number. Only rely on the result once the accuracy is comfortably smaller than the distance that matters to you, such as your distance to a county line.
6. Fall back to a manual pin. If you are on a desktop or the fix stays poor, place the point yourself on the latitude and longitude map or search for your address instead.

Location access requires a secure (HTTPS) page, and you can revoke it at any time in your browser's site settings.

Any lookup that starts from your position inherits its uncertainty. The table below is a rough guide to what each level of accuracy can support:

For example, What County Am I In shows your distance to the nearest county line. If that distance is smaller than your accuracy radius, the answer could go either way. The same test applies at state lines: see which state you are in and how close the state line is before relying on the result for anything state-specific. Likewise, turning coordinates into a street address with coordinates to address returns the nearest address the geocoder knows, which may be a neighbor's when the fix is coarse.