HIV dynamic model was developed to help our understanding of HIV infection and progression to AIDS. While there have been substaintial literatures on this subject starting in the 1980s, only very recently these models were applied to real data analysis. We have developed methods to apply the dynamic models to analyze clinical trials data where frequent tracking of viral level in the patients are usually not avialable. The application of these model to clinical data provides new insight in infection process. One major goal of our research is to utilize the information on viral levels shortly after initiation of anti-retroviral therapy to help evaluation of the new treatment. Up to now, the evaluation has always been based on endpoints after 48 weeks and the data in the intermediate period are not efficiently used. HIV dynamic model can provide a tool that extract information from these data.