The complicated machinery used by HIV for multiplication offers several targets for therapy and the viral enzymes appear to be particularly appropriate targets. For the HIV reverse transcriptase, this has been demonstrated by the clinical effect of zidovudine, but less toxic drugs are needed. The HIV protease is another viral enzyme against which inhibitors are likely to be developed. As the function and structure of HIV proteins are understood in more detail, it will become possible to construct more selective inhibitors. It is necessary to evaluate new HIV inhibitors in animal models to facilitate the rapid selection of suitable compounds and combinations for further clinical evaluation. It appears likely that several drugs preventing HIV replication will be required, and that these will appear during the next few years, but it is also apparent that the elimination of HIV from a patient remains a remote possibility.