To investigate the possible mechanisms through which dextran modifies capillary filtration coefficient (CFC), the effects of perfusion with a protein-free dextran solution were compared with those of perfusion with a Ringerlike solution. With the use of the isolated cat hindlimb, CFC was measured during perfusion at constant flow with three solutions, a control blood-albumin mixture, a Ringerlike solution called dialysate, and 3.7 g/dl dextran dissolved in dialysate. The solutions were warmed to 37-38 degrees C, bubbled with 95% O2-5% CO2, and contained 0.015 g/dl or more papaverine. CFC was calculated from the rate of limb weight gain following a step increase in venous pressure. Dextran perfusion increased CFC to 2.0 +/- 0.2 (SD, n = 8) times control, which was significantly less (P less than 0.001) than 3.1 +/- 0.6 (n = 8) times control previously reported for dialysate perfusion. The difference between the measured viscosity of dextran (1.35 cP) and dialysate (0.72) could account for this reduction. However, when a dialysate perfusion followed a dextran perfusion, CFC only increased to 2.3 +/- 0.4 (n = 8) times control. This value is also significantly less (P less than 0.01) than 3.1. This observation suggests 1) that dextran is retained within the transcapillary channel and 2) that dextran reduces CFC mainly by partially blocking the transcapillary channel rather than by increasing viscosity.