It depends a bit on the specific goal and configuration. Exiting out of a non Tor proxy allows you to protect yourself entirely from active attackers who target only the Tor network: at best they will be able to determine that you have accessed a non-Tor related proxy. The same benefit comes from entering with a VPN. On the other hand if you exit with a fixed proxy you will hurt your anonymity a lot probably, but you are pseudonymous anyway and as long as you are always pseudonymous with Tor and the same pseudonym is used, it shouldn't matter imo. Entering with a VPN also protects you from active attackers who only target Tor and it can give you a nice static entry node which comes with quite a lot of advantage really. On the other hand if your VPN node is compromised then you will always use a compromised entry guard. The more technically advanced people who are part of the Tor community tend to talk about it as if it is a magic shield and using anything else with it is either going to hurt your anonymity or have no effect on it. This is a very commonly held opinion in Tor circles, even amongst some of the people who know Tor best. However you also need to keep in mind that they tend to look at things a bit differently than you may. They like the fact that Tor changes entry guards and selects several, because let's say you have three entry guards and one of them is bad, assuming you use them all equally this means that the attacker has the ability to view only 33% of your entry traffic. If you use one entry node and it is bad the attacker has the ability to view 100% of your entry traffic. On the other hand if you use three entry nodes that means that there are three chances for you to pick a bad node, if you use one entry node and it is good that means that you are perfectly protected from purely active attackers. And it is much easier to pick one good entry guard than to pick three. But if you pick one and it is bad you are in a much worse situation than if you picked three. The Tor community tries to balance these things but there is some significant room for interpretation imo. They are also quite firm about researching everything extremely thoroughly, imo to the point that sometimes they delay making obvious good choices simply because there is not a paper out backing them yet. One example of this is how long it has taken them to go about switching to using two entry guards rather than three. it is not a surprise to me that this is superior, anymore than it is a surprise to me that using layered guards can greatly increase the anonymity of hidden services, but as obvious as these things appear to be only recently have research papers been released regarding them. Another example is the end point linkability issue I pointed out: Tor really does aim to be an anonymity network but the way we use it we are not anonymous anyway due to our pseudonyms. Tor people may want circuits to rotate pretty quickly to maintain unlinkability and thus increase anonymity in one sense, but conversely fast circuit rotation increases the speed in which an attacker will be able to monitor both ends of your connection (although having guard nodes ameliorates this by a lot). The fact that using a proxy to exit through will make you less anonymous is the main reason people who think in terms of Tor will strongly suggest against it, but I think they may not see that for people who don't care about true anonymity that protecting from Tor focused active attackers may be more of a plus than sacrificing their already non-existent anonymity is a minus (pseudonymity != anonymity).