To determine the reliability of creatinine as a measure of the glomerular filtration rate (GFR), we compared the simultaneous clearance of creatinine to that of three true filtration markers of graded size in 171 patients with various glomerular diseases. Using inulin (radius [rs] = 15 A) as a reference marker, we found that the fractional clearance of 99mTc-DTPA (rs = 4 A) was 1.02 +/- 0.14, while that of a 19 A rs dextran was 0.98 +/- 0.13, with neither value differing from unity. In contrast, the fractional clearance (relative to inulin) of creatinine (rs = 3 A) exceeded unity, averaging 1.64 +/- 0.05 (P less than 0.001), but could be lowered towards unity by acute blockade of tubular creatinine secretion by IV cimetidine. Cross-sectional analysis of all 171 patients revealed fractional creatinine secretion to vary inversely with GFR. This inverse relationship was confirmed also among individual patients with either deteriorating (N = 28) or remitting (N = 26) glomerular disease, who were studied longitudinally. As a result, changes in creatinine relative to inulin clearance were blunted considerably or even imperceptible. We conclude that true filtration markers with rs less than 20 A, including inulin, are unrestricted in glomerular disease, and that creatinine is hypersecreted progressively by remnant renal tubules as the disease worsens. Accordingly, attempts to use creatinine as a marker with which to evaluate or monitor glomerulopathic patients will result in gross and unpredictable overestimates of the GFR.