We combine several databases to answer what an IP address is, where it is and how risky it looks. Here’s where the data comes from and how accurate it is.
IP Geolocation
We use a composite approach: two independent geolocation databases are queried for every IP, and results are merged. MaxMind serves as the primary source; DB-IP fills any gaps. Postal codes are approximated from coordinates when not available natively. You can test the accuracy yourself with our IP Lookup tool.
| Source | Database | Update Frequency | License |
|---|---|---|---|
| MaxMind | GeoLite2 City + ASN | MaxMind publishes twice a week; we download weekly (Wednesdays) | GeoLite2 EULA |
| DB-IP | DB-IP Lite City + ASN | DB-IP publishes monthly; we download it on the 2nd | CC BY 4.0 |
| GeoNames | Postal code coordinates | Periodic | CC BY 4.0 |
Geolocation Accuracy
- Country level: the most reliable level, though not perfect: an address can still be placed in the wrong country, most often behind a VPN, a mobile gateway or a block that recently changed hands.
- State / region level: less reliable than country, and it varies by country and network.
- City level: an estimate with a radius. In the MaxMind GeoLite2 City database we use (27 October 2026 build), counted across the IPv4 address space it places (not across real lookups), 13.9% of addresses carry an accuracy radius of 10 km or less and 23.2% between 11 and 50 km, while 40.3% have a radius over 500 km. Only 45.1% of them have a city in that database. The IP lookup shows the radius when MaxMind places the address.
- Postal code: approximated from coordinates when not available natively
Accuracy is lower for VPNs, proxies, mobile networks, satellite internet, and corporate networks that route traffic through centralized gateways.
Security & Threat Intelligence
Beyond geolocation, every IP lookup checks the address against the lists and ranges below. They power our IP Reputation Check and VPN & Proxy Detector. The email spam blocklists (DNSBLs) in the last row are the exception: they are checked only when you run the IP Blacklist Check.
| Data | Source | Coverage |
|---|---|---|
| Tor exit nodes | Tor Project (official list) | ~1,300 nodes, updated every 30 minutes |
| VPN endpoints | X4BNet (open source) | ~10,000 IPv4 ranges |
| Cloud/datacenter IPs | AWS, Google Cloud, Cloudflare, X4BNet, plus known hosting networks (ASNs) | ~56,000 CIDR ranges |
| Threat IPs | IPsum (aggregated feeds) | ~43,000 IPs, updated daily |
| Attacker IPs | FireHOL IP lists, levels 1 and 2 | Updated daily |
| Public proxies (FireHOL) | FireHOL’s proxy and anonymous-network lists | Updated daily |
| Public proxies (IP2Proxy LITE) | IP2Proxy LITE (IP2Location) | Updated monthly |
| Relay IPs | Apple iCloud Private Relay | Official egress IP list, updated weekly |
| Hijacked and criminal networks | Spamhaus DROP (© The Spamhaus Project) | ~1,850 IPv4 and IPv6 networks, updated daily |
| Verified bots | The 22 IP range lists that 13 operators publish for their bots: Google, Microsoft (Bing), Apple, OpenAI, Anthropic, Perplexity, DuckDuckGo and others | ~3,500 ranges, updated daily |
| Spam blocklists (IP Blacklist Check only) | 11 DNSBLs (Spamhaus ZEN, Barracuda, SpamCop and others) | Queried when you run the tool, results kept up to 24 hours; IPv4 only |
Country Intelligence
| Data | Source | License |
|---|---|---|
| Country details (250 countries) | mledoze/countries | ODbL |
| Internet penetration & GDP per capita | World Bank Open Data | CC BY 4.0 |
| Corruption Perceptions Index | Transparency International | CC BY-ND 4.0 |
Registry and Routing Data
Every IP lookup also shows a “Registry record”: what the regional internet registry has on file for the address’s network, plus, for IPv4 addresses, the BGP prefix that network announces. It loads just after the main result. The ASN lookup uses the same registry data for each network’s abuse contact.
| Data | Source | Queries and caching |
|---|---|---|
| Network name, abuse contact, registered range and CIDR, registration and last-changed dates | RDAP, from the registry that holds the address: ARIN, RIPE NCC, APNIC, LACNIC or AFRINIC | At most one query per address looked up, to the registry IANA lists for that range (list refreshed weekly; rdap.org as a fallback); kept up to 7 days |
| Announced BGP prefix (IPv4) | Team Cymru IP to ASN Mapping | One DNS query per address looked up, kept for a day; announced addresses only |
The record names the registry that answered, for example “From the ARIN RDAP service”. When the registry does not answer within 2 seconds, the record says so; a lookup of the same address after 10 minutes asks the registry again. The location, the risk score and the IP type do not use the record.
The registry record and the BGP prefix are shown on this website only; the API does not include them.
IP Type Classification
Each IP address is classified into one of eight types based on all available signals. When several apply, the type is the first match in this order: Bogon/Reserved, Verified Bot (unless its network is on Spamhaus DROP), Tor Exit, Relay, VPN, Proxy, Datacenter, then Residential.
- Bogon/Reserved — Private, loopback, documentation, or IANA-reserved address space
- Verified Bot — Crawler, monitor or webhook sender inside the ranges its operator publishes, such as Googlebot or Bingbot
- Tor Exit — Tor network exit relay
- Relay — Apple iCloud Private Relay egress address
- VPN — Known VPN provider endpoint
- Proxy — Address on public proxy lists
- Datacenter — Hosting provider, cloud, or server
- Residential — Standard consumer ISP connection
How the risk score works
Every IP lookup on My IP Help shows a risk score from 0 to 100. It adds up the risk signals we find for the address, and the result shows each one with the points it adds and the list it comes from, so you can see why an address scored what it did.
The score is Low from 0 to 29, Medium from 30 to 69 and High from 70 to 100. It is capped at 100. What an address has done (being listed for abuse, running an open proxy or a Tor exit) counts for more than what it is (a datacenter or a VPN): Google’s public DNS, 8.8.8.8, is a datacenter address and scores Low.
| Signal | Points | Source |
|---|---|---|
| In a hijacked or criminal network | +70 | Spamhaus DROP |
| Tor exit node | +75 | Tor Project exit list |
| Listed on public blocklists | +20 for one list, +8 for each more, up to +60 | IPsum (aggregates 30+ public blocklists) |
| On FireHOL’s attacker lists | +45 (level 1), +30 (level 2) | FireHOL IP lists |
| Public proxy | +45 | FireHOL proxy and anonymous-network lists |
| Public proxy (IP2Proxy LITE) | +40 | IP2Proxy LITE |
| VPN service | +30 | VPN provider ranges |
| Datacenter or cloud address | +15 | Cloud and datacenter ranges |
| iCloud Private Relay | +5 | iCloud Private Relay ranges |
A verified bot, such as Googlebot crawling from Google’s published ranges, is not counted as a datacenter, VPN or proxy address: only the abuse lists count against it. An address in a Spamhaus DROP network is High risk even when it is a verified bot.
The lists are public or licensed. Most are updated daily; the iCloud Private Relay list is updated weekly and IP2Proxy LITE monthly.
The API leaves out IPsum and the FireHOL attacker lists, because both include DShield data, whose terms do not allow use in an API for now. The same address can therefore score lower through the API than on the website. IP2Proxy LITE is used in the API too: IP2Location confirmed in writing that the API may use it, and its credit is in the API documentation.
We do not have traffic from a network of customer sites, so the score never claims to have seen an address misbehave elsewhere. It rates an IP address, never the person using it: many people share an address, and a VPN or a datacenter address is often used for good reasons. Reserved and private addresses get no score, because they are not on the public internet.
How We Ensure Accuracy
Rather than relying on a single database, we merge two independent ones and fill the gaps they leave:
- Composite lookups: MaxMind and DB-IP are queried independently, and their results are merged with MaxMind values taking precedence. DB-IP fills gaps in timezone, postal code, and state data.
- Timezone from coordinates: When the databases have no timezone for an address, we derive the timezone directly from coordinates using a mathematical boundary lookup.
- Postal code approximation: For IPs where the database lacks postal code data, we match coordinates against a 1.8 million row GeoNames postal code dataset to find the nearest known postal code.
- Regular updates: MaxMind publishes twice a week and we download the new database every Wednesday; most threat lists update daily, and Tor exit node lists update every 30 minutes.
Known Limitations
- VPNs and proxies mask the true location. The geolocation shown will be the VPN server’s location, not the user’s.
- Mobile networks may assign IPs from a regional pool, so the city shown may differ from the user’s actual city.
- Corporate networks often route all traffic through a central gateway, so all employees may appear at the same location.
- Anycast IPs (e.g., 8.8.8.8, 1.1.1.1) resolve to the nearest server, so the reported location may vary.
- Newly allocated IPs may not appear in geolocation databases immediately. Database updates propagate within days to weeks.
See incorrect data?
If you notice inaccurate geolocation or network information for a specific IP, please let us know through our contact page. Include the IP address and what data appears incorrect. You can also use our free tools to cross-check results with DNS, WHOIS, and ASN data.