* allows catchall but will reject specific addresses (and then set it up to reject sales@, info@, contact@, accounts@, accounting@, or at least any of those that you don't actually need)
* can reject technically malformed emails.
* can reject emails depending on ip of sender mail server or address of sender.
* can reject emails if ip of sending server is on particular black lists.
* can greylist (tell the sending server to try again later) if a sender has not sent you an email before.

I use such a service and it refuses to accept a couple of hundred spam emails a day. Personally, I haven't had any false positives (that I'm aware of) but YMMV depending on how you use your email. Always think through how a particular anti-spam technique might affect your email before you implement it.

Having said that, I have been filtering a bunch of publically-posted email addresses via Gmail recently, their spam filtering is excellent and very little bad stuff gets through.

Other than that, I sue SpamAssasin on my mail servers, there are a lot of Bayesian filtering systems which are reasonably effective.