Routine collection of molecular-level measures of disease processes impacts many aspects of design and evaluation of interventions. In clinical investigations of HIV infection, this information includes number and genotype of viral particles in plasma and other body compartments as well as high-dimensional measures of immune function. Such information may allow treatment regimens (e.g. combinations of antiretroviral drugs) to be more effectively targeted to appropriate patients based on specific characteristics of virus and host. To achieve this goal, there is need for more systematic approach to screen combinations of new and available drugs and to assess effects of patient-specific characteristics on response to treatment. Studies must assess interactions among drugs, and between drugs and patient classes. The presentation will focus on recent designs and analytic approaches for classifying patients and making use of these classes in study design. Examples will be provided from studies of the AIDS Clinical Trials Group.