Three things suggest to me that gender differences are not as intrinsic as we might assume, given the current disparity:
1. Social groups tend to select for like individuals. Many women (and other under-represented people) find it surprisingly difficult to make it in the scientific, academic, and engineering communities when their advisors, peers, and bosses are overwhelmingly white males. Expectations that women "simply don't" do field X are shockingly present to this day. This suggests that we have not yet reached an equilibrium point where women are fairly represented with respect to their full potential.
2. Other countries have significantly varying gender ratios in these fields. That suggests that the difference is partly cultural.
3. In the US, we lose an astonishing number of young women from academic and engineering tracks. Standardized test performance (as well as metrics like "confidence in math ability") fall off dramatically in seventh grade--a time period where girls are under tremendous social pressure. By the time undergrad rolls around, few are still trying to become scientists, mathematicians, and engineers.
One of my former professors, Melissa Eblen-Zayas, has a good summary of links about women and minority participation in physics: http://people.carleton.edu/~meblenza/Resources.html