What the distance between two IP addresses tells you
Every public IP address belongs to a network, and every network has a place in the world where its address blocks are used. Geolocation databases record, for each block, the city it is most likely to serve. When you put two addresses side by side, the distance between their cities answers practical questions: is this login coming from where the user says they are, how far is my traffic travelling to reach that server, and is this "local" proxy really local?
The number on this page is the great-circle distance — the shortest path across the surface of the Earth — between the two reference points. It is computed with the haversine formula, which treats the planet as a sphere with a radius of 6,371 km. On real distances that error is under 0.5 percent, far smaller than the uncertainty of the geolocation itself.
How to read the result
The large figure is the distance in kilometres, with miles and the compass heading from the first address to the second underneath. Three badges summarise the comparison:
- Same country / different countries — compares the country codes from the geolocation database. Different countries are the clearest sign that a VPN or proxy is in use, or that a user is travelling.
- Same network (ASN) / different networks — compares the autonomous system numbers. Large ISPs and clouds span whole continents, so a long distance on the same ASN is normal; a short distance on different ASNs simply means two providers in one city.
- Same connection type / different connection types — compares the network classification (residential ISP, mobile carrier, hosting or datacenter, education, government). A residential address and a hosting address a few kilometres apart are typically a person and a server, not two people.
Below the badges you see each address with its city, region and country, its network name and type, and the coordinates used. The map plots both points and the straight line between them; drag or zoom to inspect it. The "full lookup" link opens the complete profile of either address with blacklist, RPKI, native-IP and purity data.
The light-time estimate
The small line under the distance gives the time light needs to cover that distance in optical fibre, one way. Light in glass travels at roughly two thirds of its speed in vacuum, about 200,000 km per second, so 1,000 km costs about 5 milliseconds. This is the physical floor for latency between the two places: no connection can be faster, and real connections are slower because cables follow coastlines and roads, traffic traverses several networks and routers, and the usual measure — a ping — is a round trip. A ping of two to three times the one-way estimate is typical; much more than that suggests a detour, for example a VPN exit on another continent or a mobile carrier that routes through a distant gateway.
Typical uses
Checking a VPN or proxy. Run the test with the first field set to the exit address your VPN gives you and the second field empty (your real connection). The distance shows how far your traffic detours, which explains added latency and whether the exit is really in the advertised city. A "New York" server that geolocates to Frankfurt is either mislabelled by the provider or relocated by the database — either way it will not pass as New York on services that check location.
Verifying a remote login or order. Support and fraud teams can compare the address of a login with the address on file. Same city or same region is expected for a regular user; a different country combined with a hosting-type network is the classic sign of an account used through a proxy.
Planning server placement. Compare the address of a server with the addresses of the users who complain about slowness. If the distance explains the latency, a server or a CDN edge closer to those users is the fix; if the distance is short and latency still high, the problem is in the route, not the geography.
Comparing two servers or two offices. Enter both addresses to get the physical separation and the latency floor between two sites, useful when planning replication or choosing which datacenter should host a shared service.
Accuracy and limits
IP geolocation is a statistical guess at city level, not a GPS fix. For fixed broadband in most countries the city is right the great majority of the time; for mobile networks the address is often mapped to the carrier's gateway city rather than the phone's location; for datacenters and VPN exits the database may record the company's registration address rather than the server's. The coordinates are the city's reference point, so two addresses in the same city show a distance of zero even if they are on opposite sides of it, and distances below about 50 km mean "roughly the same place".
Private and reserved ranges (10.0.0.0/8, 192.168.0.0/16, fc00::/7 and so on) have no location and cannot be measured. Anycast addresses such as public DNS resolvers are announced from many places at once; the database can only record one of them, so the distance to an anycast address says little about the path your packets actually take.
If a location looks wrong for an address you know, the geolocation feed is corrected monthly from the provider; the full lookup page shows the data version in use. Comparing the database location with your real position is also possible on the IP leak test, which can read your device's location in the browser with your permission and compute the same distance locally.