Since estimates in the literature for the fundamental binding parameters of dicumarol to bovine serum albumin show wide disparity, these phenomena were reinvestigated, using a nonequilibrium partitioning technique. A three-phase system was used consisting of an aqueous drug-protein phase and a separate aqueous receiving phase, both in mutual contact with a third overlying immiscible organic phase. The phases were contained in a partitioned, temperature-controlled cylindrical glass cell which could be rotated along its horizontal axis to effect gentle yet thorough mixing and drug transfer. Free (unbound) drug diffused from the protein phase sequentially into the other phases where it was analyzed. This technique allowed estimates of free and bound drug to be made over a very wide range of drug-protein ratios, using a single drug-protein solution, in a short time. Data analysis using an iterative, nonlinear, least-squares computer program indicated that there were two classes of binding sites with the following characteristics; n1 = 2.3 ± 0.15, K1 = (1.8 ± 0.23) x 105 M- 1; n2 = 14.0 ± 0.71, K2 = (3.0 ± 0.36) x 103 M− 1. A computer-solved theoretical treatment of drug transfer within the cell showed good correlation with the experimental results. The rotating cell technique possesses a flexibility in adjustment of operational parameters which should suit it to the study of the binding characteristics of various compounds.