zaur is right: in the OG books, the harder questions have higher numbers (though there is not a strict categorization - Q 50 is harder than Q 1 in that same section, but Q 50 may not be harder than Q 49).
You're right that you don't want to do only the lowest-numbered questions. How do you know what kinds of questions you can expect to get on the test? Have you taken a practice test? That should give you an idea of whether you can skip the lower-numbered questions and go for the medium and harder ones or not.
A few things to keep in mind:
(1) you MUST time yourself and keep to the time constraints; a pure percentage correct (even 100% correct) means nothing without the timing data
(2) it's better to do a random mix of questions of different types (eg, DS and PS) and difficulty levels (eg, Q30, Q41, Q53, Q82) than just a straight list of 5 Qs in a row. The real test is going to mix things up.
(3) in terms of gauging your performance, take a practice test; stats on non-adaptive banks of printed questions... not as useful
in terms of gauging your overall performance(4) stop worrying about difficulty level so much. :) Read this:
http://www.beatthegmat.com/mba/2010/03/ ... n-the-gmatFinally, it sound (from what you wrote) like your focus is mostly on doing lots of problems. It's good to do practice problems, of course, but when actually doing problems, you're learning only about 20% of what you could learn overall from that problem. Most of your learning comes from your review and analysis after the fact, so make sure that you're doing that as well. Here are some articles that show you how to do this:
http://www.manhattangmat.com/articles/a ... roblem.cfmRead the above first, then:
SC:
http://www.manhattangmat.com/articles/GMATprep-SC.cfmCR:
http://www.manhattangmat.com/articles/CR-assumption.cfmRC:
http://www.manhattangmat.com/articles/a ... estion.cfmGood luck!