SAS surpasses many packages because of its data handling capability and the many models one may easily estimate with it. In addition to the basics, one may estimate log-linear models, logit models, complementary log-log models, and models for event history data. Using Proc Mixed one may estimate multi-level models for interval-level variables, dichotomous variables, ordinal variables, count data, and more.

Stata is also a widely used program that runs on several platforms. Further, several users have written macros that can be obtained and that extend the already extensive capabilities of the program.

The Centre for Multi-Level Modeling site maintains an on-going discussion of some of the issues involved in estimating and interpreting multi-level models. Although the discussion focuses on MLWiN, I have never used MLWiN and so I cannot say much about it, except that many analysts swear by it. I am told that the software has advanced data handling capabilities. One limitation, though, is that as far as I know it is only available for PCs running Windows; no Unix version is available.

StatLib contains links to software packages, macros, and discussions of statistical methods.