First you have to honestly assess whether you've really mastered common undergraduate level topics to the level required for the GRE. If you want to test your mechanics ability, say, open Halliday and Resnick and from the first 14-15 chapters pick several two or three star problems randomly. Can you do them in a short amount of time, or do you need a long time to ponder over them, look up formulas and eventually you can do them? If the case is the latter, then I'd say you haven't had enough practice. Mend that. Do as many practice problems as you can find on topics you find that slow you down.
People keep saying that doing the past tests is the best way to prepare for the test. I can honestly say that that's not true in my experience, mainly because 3 of those tests (1986, 1992, 1996) are terribly outdated. Plus as you say, you've exhausted them at this point. Studying them further will be of little use. The best approach I think is to identify your weak topics (which you should be aware of, having taken the test twice), open the corresponding Halliday and Resnick chapters, read up on them, go through the examples, and most importantly do many many problems. In this way you expose yourself to many different types of problems and when those types of problems show up on the actual test, you'll be able to do them much faster.
You say you have trouble with circuits (don't worry, I hated them as well but with practice I've learned to get by). Use this approach to work on them. Most GRE's have at least 2-3 (sometimes as much as 5) problems related to circuits and if you can get those that's an easy 20-30 point improvement. Ditto for other topics.
Let me end with some anectodal statistics. Everyone I know who went through all of Resnick (at least 5 people) and did the majority of the problems got perfect or near-perfect scores. Most people who didn't (including myself), didn't. In contrast to what TakeruK says, I strongly believe that putting in the time to try a different approach which could improve your score is very worth it, especially if you want to apply for theory. The only case when it's probably not worth is if you're applying to astronomy programs.