A comparison of venous plasma, capillary plasma, and saliva pharmacokinetics of artemisinin was performed in four healthy subjects given a single 500 mg dose. Artemisinin was determined to be 88% bound in venous plasma. Saliva levels were more closely related with (unbound) capillary than with venous plasma concentrations. Matrix comparisons demonstrated artemisinin saliva/plasma and capillary/venous concentration ratios of well over one initially, which decreased with time to stabilize after a 3-hr period. Following this stabilization period, differences in capillary and saliva (relative to venous) levels were independent of both drug concentration and time. These observations, together with improved correlations of concentrations measured 3 hr following drug intake compared with correlations for the entire data set, indicate a putative arterial-venous concentration difference for the drug. The capillary and saliva matrices proved promising replacements for venous sampling in the quantitation of artemisinin concentrations. Due to their relative ease of obtainment and a greater patient acceptability, capillary and saliva sampling may be of particular value in field-based and pediatric pharmacokinetic studies with artemisinin.