Indinavir and nelfinavir plasma levels were studied in 407 patients having plasma HIV RNA <500 copies/mL after 4 months of treatment with these drugs. For each drug, an observed/predicted (O/P) ratio was calculated between individual and mean time-adjusted population plasma drug levels. The relationship between the O/P ratio and the risk of rebound of plasma HIV RNA >500 copies/mL beyond month 4 was studied using Cox proportional hazard models. Median follow-up was 20 months. There was no association between indinavir plasma levels and risk of virologic rebound, whereas low nelfinavir + M8 (active nelfinavir metabolite) plasma levels were associated with a higher risk of virologic rebound. In multivariate analysis, the adjusted relative hazard of virologic rebound for patients with an O/P ratio of nelfinavir + M8 metabolite <0.8 compared with others was 2.2 (P = 0.01). In some patients, plasma levels of nelfinavir sufficient to achieve early viral response may not be sufficient to maintain it in the long term. This may be related to insufficient compliance with dietary recommendations. Monitoring of nelfinavir plasma levels thus seems useful, even in patients having early virologic response.