Pharmacokinetic studies rely on blood sampling at times relative to predefined dosing intervals. Intensive sampling is often done under direct observation of dose taking, which, though costly, virtually eliminates uncertainty about actual dosing times. In contrast, the sparse sampling done in population pharmacokinetic studies relies on patient-reported times of dosing, the accuracy of which the authors sought to assess by adding electronic monitoring to the usual patient reporting of dosing times. The study involved 35 antiretroviral-naive, human immunodeficency virus-infected patients and was designed to assess the safety, tolerability, pharmacokinetics, and antiviral activity of prescribed lopinavir/ritonavir (800/200 mg qd or 400/100 mg bid), stavudine, and lamivudine. The present research reports the pharmacokinetic analysis that results from taking into account the patients' actual dosing histories. Intensive sampling for plasma lopinavir concentrations was done at week 3, and 4 additional predose (trough) concentrations were measured during the next 12 months. Convergence was achieved by fitting a simple 1-compartment pharmacokinetic model, with first-order absorption and elimination, to the sparse sampling data, using electronic monitoring-reported times. In contrast, convergence was not achieved using the simple model when steady state was assumed, and the times for the last qd dose or the last 2 bid doses, as reported by the patient, were used as model input. Estimated individual pharmacokinetic parameters were then combined with electronic dosing histories to project each patient's internal drug exposure over long periods of time. This strategy may provide a basis for greatly increasing the informational yield and utility of conventional therapeutic drug monitoring.