Efavirenz is a drug subject to extensive metabolism, mainly by the cytochrome P-450 isoenzyme CYP2B6, known to exhibit extensive interindividual variability. The aim of the present study was 2-fold: to investigate the relationship between plasma concentration and clinical effects of efavirenz and to investigate the extent of the inter- and intraindividual variability of the plasma concentration measurements. From an open clinic, 68 HIV-positive patients on efavirenz-containing treatment were recruited. From each patient 1 to 5 samples were collected; 43 had more than 1 sample taken. Most samples were taken 10–24 hours after the latest dose. Efavirenz was analyzed by high-performance liquid chromatography with UV detection. The data were analyzed by the variance component model analysis of variance. Efavirenz concentrations were reproducible, and intraindividual variability constituted only 16% of the total variance. Thus, 84% of the variance was attributed to interindividual variability. The incidence of primary treatment failure was related to low plasma concentrations with a geometric mean concentration of 6.1 μmol/L compared with 8.7 μmol/L in those responding to therapy (P < 0.05). If a cutoff of 7 μmol/L is used, 10 of 13 failing to respond were below this level compared with 15 of 45 in those responding. It is concluded that efavirenz plasma concentration measurement gives reproducible results predictive of primary treatment failure. A lower bound for the therapeutic level of 7 μmol/L is proposed, and data from other authors suggests that an upper level of 13 μmol/L may be applied.