Partition coefficients, k p , of chlorpromazine between the aqueous phase and lipid bilayer vesicles were determined as function of drug concentration, lipid chain length, cholesterol content and temperature encompassing the range of the lipid phase transition. Radioactivity and absorption measurements were performed to determine the k p values. Up to a concentration of 3·10 −5 M, the partition coefficient is independent of chlorpromazine concentration, whereas it decreases drastically at higher chlorpromazine concentrations, at which membrane lysis is observed. Membrane structure is not disturbed at less than 3·10 −5 M chlorpromazine, as was concluded from electron paramagnetic resonance studies measuring TEMPO partitioning and order degree. However, the lipid phase-transition temperature decreases and is broadened at higher chlorpromazine concentrations. From fluorescence measurements, we conclude the formation of chlorpromazine micelles at concentrations higher than 5·10 −5 M in chlorpromazine in the absence of lipids and the formation of mixed micelles in the presence of lipids. The effect of lipid chain length on k p values was investigated. The partition coefficient decreases from 8100 in dilauroyl- to 3400 in dipalmitoylphosphatidylcholine vesicles, both at 50°C, that is, above their corresponding phase-transition temperature t t . At t < t t the k p values are strongly reduced, by at least a factor of 10, depending on lipid chain length and membrane composition. It is possible to establish a lipid phase-transition curve from the temperature-dependent measurements of the k p values. Cholesterol within the lipid membrane strongly decreases k p . At 20 mol% cholesterol in dipalmitoylphosphatidylcholine membranes, the partition coefficient is reduced from 3400 to 2300. This value is well comparable to the k p value obtained in erythrocyte ghosts. In contradiction to earlier experiments by Conrad and Singer (Biochemistry 20 (1981) 808–818), this value in a biological membrane could be obtained by the hygroscopic desorption as well as the centrifugation method. From our experiments we are justified in further considering artificial bilayer membranes as models for biological membranes.