The convective component in the general description of transport of solutes across the peritoneal membrane can be expressed as SQuc, where S is the sieving coefficient, Qu is the ultrafiltration flow rate, and c is the average concentration in the membrane (c = (1-F)cB + FcD, where cB and cD are blood plasma and dialysate solute concentration, respectively). F is a weighing function dependent on Qu, S, and the diffusive mass transport coefficient KBD. In this study a class of simple models of solute transport was considered in which S = 1 (justified for small solutes) was chosen, and F was selected as follows: F = 0 (as in the S = 1 (justified for small solutes) was chosen, and F was selected as follows: F = 0 (as in the widely used model of Garred and coworkers), F = 0.5 (theoretically justified model), F = 0.33 (theoretically justified for a high ultrafiltration period), and F = 1 (for convective transport from dialysate to blood). For all these models the estimation of KBD from clinical data can be performed with the aid of linear regression. The simple models were compared with the Pyle-Popovich model which takes into account the general expression for convective solute transport, for both the accuracy of the KBD determination (using linear regression) and the accuracy of theoretically calculated dialysate to plasma concentration ratios (D/P) to experimental D/P. Clinical evaluation of the new models was carried out in 28 6-hour dwell studies in 21 nondiabetic patients using 2 liters of hypertonic (glucose 3.86%) dialysis fluid. The differences between the simple models were small from the clinical point of view for urea, creatinine, glucose, and potassium, whereas for sodium the predictions were not satisfactory for any of the models. For urea and creatinine the model with F = 0.5 yielded the best fit of theoretical predictions to experimental data. For glucose and potassium small but systematic deviations of theoretical D/P from experimental D/P were observed for all simple models. The protein transport could be satisfactorily described by a model in which F = 1, as shown for total protein.