To study the respective roles of indinavir concentrations and treatment adherence as predictors of early virologic response, we analyzed the patients of the APROCO cohort treated by indinavir 800 mg TID during the first 4 months. Minimum (Cmin), maximum (Cmax), and the ratio of the measured to expected concentrations (CR) were estimated for each patient at M4, from a population pharmacokinetic analysis of all data. The relationship among virologic success at M4 [plasma HIV RNA (VL) <500 copies/mL], baseline characteristics, estimated indinavir concentrations, and adherence score measured by a self-administered questionnaire, was analyzed by multivariate logistic regression. In the 216 studied patients, baseline median HIV RNA was 4.4 log10 copies/mL, and CD4 cell count was 309/mm. Virologic success was achieved in 195 (90%) patients; it was independently related to baseline viral load (OR = 0.524, CI 0.29-0.93; P = 0.03), antiretroviral treatment naive status (OR = 3.89, CI 1.29-11.76; P = 0.01), and indinavir Cmin (OR = 1.06, CI 1.02-1.10; P = 0.004) when adherence score was not included in the model, whereas full adherence was the only independent related factor when included in the model (OR = 8.8, 95% CI 2.85-27.3; P < 10). In the 168 fully adherent patients, virologic success was more frequent in patients with shorter duration of antiretrovirals at baseline (P = 0.03), lower baseline HIV RNA (P = 0.03), and higher indinavir CR (P < 10); the most discriminating Cmin cut-off was 194 ng/mL. Data on the relationship between indinavir plasma concentration and virologic success are therefore misleading without a concomitant assessment of adherence. These data suggest that any strategy of therapeutic drug monitoring must imply first a combined evaluation of plasma concentrations and adherence level and second an intervention target based on the results of both assessments.