Skip to content
InternetData

Is it safe to block mobile proxy IPs?

Mobile proxy detection, measured: 70.5% of mobile proxy IPv4 addresses are seen on one day, and widening a hit to its /24 blocks 130 times as many.

Published 4 min read

On this page
  1. Most are seen for a day
  2. What yesterday's lists caught
  3. Where they cluster
  4. A rule for mobile proxy hits
  5. Mobile proxies are in the residential list too
  6. How we measured

Proxy sellers have a ready answer for why their mobile IPs are hard to block: carrier-grade NAT puts many phones behind one public IPv4 address, so a block on the proxy is a block on the carrier's customers too. The sharing is real. A study of a large CDN's traffic published in August, Hsu, Pearce, Li, Berger and Richter's Detecting and Characterizing Massively Shared IP Addresses, describes carrier-grade NATs that "share hundreds or thousands of users on a single IP", and found "over 40% of total traffic coming from less than 2% of active IP addresses".

What that argument leaves out is time. A mobile proxy holds a carrier address for a while, then the address goes back to the carrier's pool. So we measured how long. Our Mobile Proxy IP database's 90-day list, as published on 2026-10-09, holds 2,497,210 IPv4 addresses seen as mobile proxies between July 11 and October 8. 70.5% of them were seen on one day only.

Most are seen for a day

Days from first to last sightingIPv4 addressesShare
Same day1,759,68170.5%
1 to 6264,73310.6%
7 to 29213,5758.6%
30 or more259,22110.4%

That's when we saw each address, which is all a list can go on: an address seen once may have carried proxy traffic for longer.

IPv6 turns over faster: 92.1% of the list's 1,015,034 IPv6 addresses were seen on one day. Carrier NAT is mostly an IPv4 question, since a phone on IPv6 normally gets a prefix of its own, though the study above found "some surprising exceptions among mobile providers". The rest of this post is IPv4.

What yesterday's lists caught

The 1-day list published on 2026-10-09 holds the 50,092 IPv4 addresses seen as mobile proxies on October 8. Each list published a day earlier, on October 8, already held this many of them:

Mobile Proxy IP list, 2026-10-08IPv4 in itOf October 8's 50,092Share of the list seen again
1-day65,5015,341 (10.7%)8.2%
7-day387,95314,949 (29.8%)3.9%
14-day688,51017,858 (35.7%)2.6%
30-day1,168,33620,933 (41.8%)1.8%
60-day1,761,34524,512 (48.9%)1.4%
90-day2,503,45026,906 (53.7%)1.1%

Nearly half of the day's mobile proxy addresses weren't in the 90-day list at all. For residential proxies, the day before's 90-day list held 91.4% of the day's IPv4 when we checked on 2026-09-27.

The 90-day list holds 38 times the addresses of the 1-day list and caught 5 times as many of the next day's. 1.1% of its addresses turned up as mobile proxies again on October 8.

Where they cluster

The 90-day list's 2,497,210 addresses sit in 251,039 /24 blocks. The median block holds two of them, and 42.8% of blocks hold one. But most addresses are in dense blocks: 81.4% sit in the 57,023 blocks holding 10 or more, and 24.3% in the 3,892 holding 100 or more. 752 blocks hold 200 or more.

That's where widening a hit to its /24 goes wrong. October 8's 50,092 addresses sit in 25,549 /24s, 6,540,544 addresses in all: 130 times as many. 1,114,338 of those, 17.0%, were seen as mobile proxies at any time in the 90 days to October 8. The other 5,426,206 never were.

A rule for mobile proxy hits

  • Block, or step up to a challenge, on the 1-day and 7-day lists. A hit there is at most a week old.
  • Score on the 30 to 90-day lists. Each address carries first_seen, last_seen, hits and hits_days_pct (schema), so a hit from last week can count for more than one from July.
  • Match the address, never its /24.

Mobile proxies are in the residential list too

2,338,823 of the 2,497,210 mobile proxy IPv4 addresses, 93.7%, are also in the 90-day list of our Residential Proxy IP database from the same day. A team that already checks residential proxies flags most mobile ones. What Mobile Proxy IP adds is which of those hits are carrier-issued addresses, the ones most likely to sit behind carrier NAT.

How we measured

Every figure is from the Mobile Proxy IP lists published on 2026-10-09, each window ending October 8, except the check against the lists published a day earlier. An address counts as seen on one day when its first_seen and last_seen fall on the same date. A /24 is the 256 addresses sharing an address's first three octets. The residential match is on the full address, against Residential Proxy IP's 90-day list from 2026-10-09. On each day from August 10 to 16, the check against the day before's lists gave 18 to 20% for the 1-day list and 54 to 57% for the 90-day one.

To license the databases, see pricing or talk to us.

Written by

  • Uman Shahzad

    Founder & CEO

    IP intelligence data expert for over 8 years, former lead engineer at IPinfo, Uman has pioneered many systems and practices used by IP data companies to operate at scale, provide high-quality data and offer the best developer experiences.

Get new posts by email

Or follow along in a feed reader.

RSS feed

Put this data to work.

Everything these posts report on comes from the databases we license, and you can test them first.

00 · Evaluate

Test the real build first.

An evaluation license puts a full current build in your hands — every row, every column, no sampling — so you can measure it against your own traffic before anyone talks about terms.

01 · License

Then license what you use.

Standard, for detection and decisions inside your own product and operations.
Redistribution, when the data ships onward inside what you sell — priced by scope, not by seat.
Annual terms, a named contact, and flexible payment terms — monthly, quarterly or yearly.