[3H]Palmitate, PA, exchange efflux kinetics is recorded from human erythrocyte ghosts to buffer with bovine serum albumin, BSA, at 0 degrees C. The effects have been investigated of three medium/ghost volume ratios: 36, 80 and 500, of six BSA concentrations, [BSA]: 0.01, 0.02, 0.05, 0.2, 1 and 2% (1.5, 3.0, 7.5, 30, 150 and 300 microM) and of various v, molar ratios of palmitate to BSA, between 0.15 and 0.94. Data are analyzed in terms of a virtually closed three-compartment model. In theory, the tracer efflux is biexponential and the rate coefficients differ at least 20 fold [1]. The efflux rate at 2% BSA is monoexponential beyond our resolution time of about 1 s, but nearly biexponential at or below 0.2% BSA with a well-defined smallest-rate coefficient beta. beta depends strongly on [BSA] but is remarkably v independent. The medium/ghost volume ratio has no effect on beta when [BSA] > or = 0.2%, although beta measured at 2% BSA is almost 2-fold higher than at 0.2%. This suggests the presence of an unstirred layer, USL. According to our model, the observations are understood quantitatively on basis of our previously published dissociation rate constants of the PA-BSA complex, as well as PA equilibrium bindings to ghost membranes (Bojesen, I.N. and Bojesen, E. (1991) Biochim. Biophys. Acta 1069, 297-307). Essentially, beta is theoretically a function of two terms, one comprising the membrane transport parameters and the other the medium-dependent variables. Most important is the clearance with respect to monomer concentration adjacent to the membrane. The clearance is calculated on basis of quasi-stationary diffusion in USL. The data are compatible with a planar USL of 6 microns depth and with the same area as a ghost but not with a spherical USL.