The level of the trypanocidal drug melarsoprol was determined in serum and cerebrospinal fluid (CSF) of six healthy vervet monkeys after intravenous application of the drug following a standard treatment schedule and a recently suggested alternative protocol. The maximum serum levels measured were about 3 micrograms/ml. A three-compartment model was used to analyze the serum data. The mean residence time calculated for melarsoprol in serum was 18 h, the volume of distribution was 3.6 l/kg and the clearance was 3.5 ml/min*kg. In the CSF the drug levels were generally very low, not exceeding 55 ng/ml, and the adaptation of the drug levels was found to be very low. The comparison of the drug concentrations required to eliminate trypanosomes in vitro and the drug concentrations reached in the CSF during treatment revealed that the latter might be insufficient in some cases to eliminate all trypanosomes from this site. The peak serum levels during alternative application of the drug were lower compared to those during empirical treatment. No evidence for drug cumulation in the body was found. The results of this study are compared with recent pharmacokinetic data from human patients, and discussed in the context of the problem of relapses and reactive encephalopathy occurring after treatment of sleeping sickness.