The HIV positive selection mutation database is a large-scale database available at http://www.bioinformatics.ucla.edu/HIV/ that provides detailed selection pressure maps of HIV protease and reverse transcriptase, both of which are molecular targets of antiretroviral therapy. This database makes available for the first time a very large HIV sequence dataset (sequences from approximately 50 000 clinical AIDS samples, generously contributed by Specialty Laboratories, Inc.), which makes possible high-resolution selection pressure mapping. It provides information about not only the selection pressure on individual sites but also how selection pressure at one site is affected by mutations on other sites. It also includes datasets from other public databases, namely the Stanford HIV database [S. Y. Rhee, M. J. Gonzales, R. Kantor, B. J. Betts, J. Ravela and R. W. Shafer (2003) Nucleic Acids Res., 31, 298-303]. Comparison between these datasets in the database enables cross-validation with independent datasets and also specific evaluation of the effect of drug treatment.