I realize that the math toolkit library review ended on 4/27 but better late than never. I have been experimenting with it for the last two weeks. In particular, I have tried to use several distributions (normal, student t, uniform etc), some special functions, and root finders. In all my tests the math library performed as expected, it was fast and reasonably well documented. Note that my testing did not include all functionality in the library. One suggestion - the "normal_distribution" which is implemented by the math library puts it in the boost::math namespace. The "normal_distribution" random generator which is implemented by the random library puts it in the boost namespace. In order to be consistent (and avoid conflicts) the random library should put its classes in the boost::random namespace. All in all, I find the library very useful and I already started using it in my code. Vladimir Valenta