I have worked in at least 7+ different companies, both large and small. I have had enforceable code reviews of my code in, count them, exactly 1(!) job. I have attempted to institute them in two others, with varying degrees of success. At my current position, what I have been doing is exactly what kvale suggests - I ask my peers to review my code. That serves several purposes:

1. I get to verify that my code is good.
2. As I'm the strongest Perl developer, they get to see what I consider to be best practices.
3. It starts them thinking that code reviews might be ok.

Now, part of the problem I have had in most places is that management has never felt code reviews to be worthwhile. Say all you want about grassroots efforts, and a lot of it is relevant. But, unless some manager (the higher, the better) indicates that code quality is a priority, then code reviews (and testing, for that matter) will never achieve the level of importance that they deserve.

Now, one thing you can use in an academic setting to help prod your peers into code reviews is the threat that bugs will taint their papers. An academic's reputation is only as good as the last paper published. If that paper is disproven because of a coding error ... ouch! Kinda like you don't want to be the guy that drew the design backwards or confused metric with english measurement.

So, once you get them worried about their code quality, you pull out McConnell's books and show them the single best defense against bugs in new code - code reviews. And then you demonstrate how code reviews leads to the single best defense against bugs overall - code reuse, which also helps reduce the time taken writing the new programs. (This last point is especially important for academics because coding is generally the last thing they want to be doing.)