Partial adherence with a prescribed or randomly assigned dose gives rise to unintended variability in actual drug exposure in clinical practice and during clinical trials. There are tremendous costs associated with incomplete and/or improper drug intake-to both individual patients and society as a whole. Methodology for quantifying the relation between adherence, exposure and drug response is an area of active research. Modeling and statistical approaches have been useful in evaluating the impact of adherence on therapeutics and in addressing the challenges of confounding and measurement error which arise in this context. This paper reviews quantitative approaches to using adherence information in improving therapeutics. It draws heavily on applications in the area of HIV pharmacology.