In this study the osmotic barrier characteristics of the rat peritoneal membrane were investigated. Fluid movements between the peritoneal cavity and the blood were measured following instillation of isotonic saline (control) and hypertonic solutions of NaCl, glucose, sucrose, raffinose and myoglobin (test solutions). Moreover, 5 and 8% albumin in NaCl were investigated. Osmotic transients were assessed using a simple volume recovery technique. Peritoneal osmotic conductances (i.e. products of peritoneal hydraulic conductances [LpS] and solute reflection coefficients [sigma]) were calculated from the differences in the rates of peritoneal fluid loss and in osmotic pressures between test solutions and the isotonic saline control solution. The osmotic conductance to glucose was estimated to be 1.63 microliters min-1 mmHg-1 m-2 and that for albumin to be 59.6 microliters min-1 mmHg-1 m-2. Assuming an albumin sigma of 0.9, the sigma of glucose was estimated to be 0.025, in accordance with previous measurements for the cat peritoneal membrane. The osmotic conductances assessed here were compatible with an 'overall' peritoneal equivalent small pore radius of 47-48 A, but could also be fitted to a three-pore model of peritoneal permselectivity, including a transcellular (ultra-small pore) pathway and a large pore pathway. The great discrepancy between peritoneal sigma for small solutes and that for albumin obtained in this study indicates that small solute reflection coefficients are close to zero while that for albumin is not far from unity. Furthermore, the peritoneal hydraulic conductance (ultrafiltration coefficient) is large enough to allow for a substantial absorption of fluid directly into the plasma when the crystalloid osmotic pressures in blood and peritoneal dialysate are in equilibrium.