The human immunodeficiency virus HIV which causes AIDS poses a grand challenge to medicine. By evolving inside the patient the virus secures its survival in the presence of evolutionary pressure by the patient’s immune system and by applied drug therapy. The high evolutionary dynamics of HIV results in an essentially unique virus population for each patient. Over two dozen antiviral drugs are available whose combinations affords many hundred plausible therapy options out of which one has to be selected that is optimal or at least promising. This combinatorial problem can be effectively solved with bioinformatics support. At the center of the approach are state-of-the-art statistical learning methods which are applied to the relevant clinical and virological data sets. The talk charts the basic scenario, reports on the clinical experience with existing bioinformatics tools, presents new developments and comments on how HIV bioinformatics could be a model for personalized therapy for other diseases.