Variations in plasma protein binding may have profound effects on both disposition and activity of drugs, especially for those which are tightly bound to proteins, such as anticancer platinum derivatives. Methods of separation of the non-protein-bound fraction and some technical parameters may influence the results. We have compared ultrafiltration and ultracentrifugation, as well as the effect of potentially interfering factors, upon the determination of the plasma unbound platinum fraction after oxaliplatin administration to cancer patients. Ultrafiltration and ultracentrifugation provided very closely correlated results, so that either technique can be used. The ultrafiltration cut-off (3000-30,000 Da) devices, the type of tube for blood sampling and the type of anticoagulant (none, lithium heparinate or EDTA) did not influence the results markedly. In contrast, results were greatly influenced by freezing: erratic results were obtained on thawed plasmas when compared with those on fresh serum or plasma. Consequences may be important in usual practice, since many pharmacokinetic studies are carried out in multicentric trials with plasma processing centralized in one reference laboratory. The methods for the determination of protein-drug binding should be standardized and guidelines elaborated where optimal conditions for each type of binding assay are given.