OBJECTIVE Differential exposure to saquinavir/ritonavir may lead to therapy failure. The objective was to identify factors that influence variability of saquinavir/ritonavir plasma concentrations. METHODS Saquinavir/ritonavir data, dosed as 1600/100 mg once daily, from three separate pharmacokinetic studies, in 45 patients from Thailand and the UK, were pooled. Pharmacokinetic parameters were based on non-compartmental analysis. Univariate analysis was performed with saquinavir as the dependent variable, and ritonavir area under the curve (AUC), gender, body weight, body mass index (BMI) and study site as independent variables. Variables with a P value <0.10 were included in a multivariate linear regression analysis. RESULTS Higher saquinavir AUCs, maximum concentrations (Cmax) and minimum concentrations (Cmin) were seen in Thai patients than in UK patients. Univariate analysis showed associations between body weight, gender, study site and ritonavir AUC and saquinavir AUC (P < 0.05), whereas BMI (P = 0.13) did not. In the multivariate analysis, ritonavir AUC (P = 0.0001) and study site (P = 0.0021) were significantly related to saquinavir AUC (R2 = 0.50). CONCLUSIONS The ritonavir AUC and study site appeared to be related to exposure of saquinavir. Study site should be viewed as the total of country- and study-specific differences--such as differences in lifestyle, environment, genetic background and dietary composition--between the analysed studies.