Osmotic reflection coefficients (sigma) for a variety of solutes ranging from NaCl to albumin were determined in perfused maximally vasodilated rat hindquarters employing the osmotic transient method (Vargas & Johnson 1964). Measurements were performed at high flows and using short tubings with small volumes. Intracapillary solute concentrations of the osmotic transients were measured or estimated for solutes of the size of inulin or smaller. The PS for Cr-EDTA and cyanocobalamine were determined repeatedly in half of the experiments using an on-line modification of the single injection (indicator diffusion) method (Rippe & Stage 1978) and capillary filtration coefficients (CFC or LpS) were followed in all experiments. The capillary osmotic reflection coefficient was determined to 0.05 for NaCl, to 0.08 for sucrose, to 0.39 for inulin, to 0.57 for myoglobin and to 0.87 for albumin. These reflection coefficients were compatible with a 'small pore radius' of approximately 40 A (slit width (w) of approximately 50 A) according to modern hydrodynamic theories for the reflection coefficient and the parallel transcapillary pathway hypothesis. The best fit of the osmotic transient data to current theories for the reflection coefficient occurred if the major portion (86-87%) of the hydraulic conductivity (Lp) was accounted for by this paracellular 'small pore' (slit) pathway and if 3.0-4.1% of Lp could be ascribed to a transcellular pathway (sigma approximately I) while the remaining fraction (10%) of Lp was accounted for by a non-selective paracellular pathway (sigma approximately o); that is, by 'large pores'.