The multiple-indicator-dilution (MID) technique was used to separate solute flux (Js) across the glomerular and postglomerular capillaries in vivo. Anesthetized mongrel dogs (n = 20) during mannitol diuresis received a pulse of extracellular indicators including 125I-albumin (plasma reference), [14C]inulin (glomerular reference), creatinine (interstitial reference), and a neutral [3H]dextran (specific mol wt between 10,000 and 24,000 dalton) in the left renal artery. Left renal venous and ureteric outflows were rapidly sampled. 3H-labeled dextrans 10,000-15,500 had renal vein mean transit times (t) greater than those of 125I-albumin, indicating postglomerular extraction. 3H-labeled dextrans greater than 15,500 had t values identical to those of 125I-albumin, indicating only unidirectional glomerular extraction. The glomerular fractional dextran extractions relative to simultaneously injected [14C]inulin (ED/Ei) were calculated from urine and renal vein outflow curves and ranged from 0.98 +/- 0.02 to 0.33 +/- 0.12 (SD) for mol wt 10,000 +/- 24,000. ED/Ei values were quantitatively identical to constant-infusion fractional clearances of the same dextrans. Renal plasma flow (RPF) was then deliberately reduced two-to threefold in the same dog. ED/Ei as measured by the MID technique remained unchanged during graded reduction in RPF. In constant-infusion experiments RPF was reduced from 5.78 to 2.77 ml X s-1 X 100 g-1 and GFR from 50.4 to 36.3 ml X min-1, but the fractional neutral dextran clearances remained constant. The predominance of convective solute flux across the dog glomerulus permitted calculation of glomerular reflection coefficients 0.03 +/- 0.06 to 0.85 +/- 0.03 (SD) for neutral 3H-labeled dextrans 10,000-24,000 dalton.