I have read messages from some users stating that the first of three tor nodes is fixed constant and that this is intentional. My question here is about when this changed in the version numbering because I have seen posts about this question from summer 2015 and I only installed Tor in Jan 2016....

Whenever I request a "new tor circuit for this site" it always sets the first IP the same (to an IP in the USA). This was not like that just a few weeks ago. Current Tor browser version is 5.5.5. My original installation version is given as 5.0.7 (installed since Jan 2016). I don't understand why this would change and considering previous postings saying this was due to a Tor version change were posted in Jun/Sep 2015 last year how come versions I have used since January 2016 did not always work this way? By the way this has persisted despite me demanding around 50 new Tor circuits yesterday and also today (28/5/16) the same - always the same first IP address without fail and I have tested this for .com and .co.uk and .de websites, no change.

Also, how can I make sure this first fixed IP is trustworthy/genuine?

2 Answers 2


The length of time that you keep the same Entry Guard is called the rotation period. It's currently set to 2 months.

The reason for doing this is to try to prevent "predecessor attacks", which are discussed in this previous question, and perhaps most succinctly explained here.

With regards to when this was introduced, certainly before 2016. It looks like Ticket 8240 is a good candidate, which is still marked open, but which has had commits made under it. I think this commit in March 2013 is the culprit. I'm unsure how to explain the behaviour you were seeing.

(There's a further open ticket which is proposing raising the rotation period to an even larger value, perhaps as much as 9/10 months.)

If you'd like to learn more, see Changing of the Guards and One Fast Guard for Life.

Also, how can I make sure this first fixed IP is trustworthy/genuine?

You can't, and you should assume that it possibly isn't. But this isn't necessarily a problem. At the point your traffic hits the entry guard it has 3 layers of (Tor) encryption, and more likely than not (HTTPS) TLS encryption. One potential problem would be if the same adversary also owned your exit node, at which point they could correlate your traffic entering and leaving the Tor network in a correlation/timing attack.


Thanks again Richard.

By the way, in case anyone finds need for it (copying advice found elsewhere) I found that I could control the Entry and/or Exit node country (codes GB, FR, DE, etc) by inserting some lines within the torrc-defaults file found in the .\Browser\TorBrowser\Data\Tor subdirectory of the installation (on Windoze, don't ask me why). There is a comment warning in this file that it shouldn't be changed but I figure its free software (as in freedom) so if I want to then I can do so anyway, right? (Disobey!!!)

It is possible to insert one or two lines after:

"# If non-zero, try to write to disk less frequently than we would otherwise.
AvoidDiskWrites 1"

such that it reads:

"# If non-zero, try to write to disk less frequently than we would otherwise.
AvoidDiskWrites 1
ExitNodes {GB}
EntryNodes {FR}"

The first IP of the three remains constant (in France for this case). The third IP becomes a GB one in this case but will change (within GB in this case). If I remove the ExitNodes and EntryNodes statements restoring the file to its original content then exit and restart Tor Browser I find that the last set EntryNode remains fixed to the country and indeed IP address which arose previously (...not a US one wahay, no offence to world-wide-beloved Americans such as Bernie Sanders and Ed Snowdon, nor being so naive to imagine US servers have US IP addresses of course... just testing y'know).

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .