OBJECTIVE To compare the urinary excretion of beta 2-glycoprotein-1 with that of two other markers of early tubular disorder in diabetic patients without clinical proteinuria. METHODS The urinary excretion of retinol binding protein, beta 2-glycoprotein-1, and N-acetyl-beta-D-glucosaminidase was measured in 90 known diabetic patients who had a negative reagent strip test for proteinuria. RESULTS Among 43 patients with urinary albumin excretion within the reference range, 23 (53%) had raised urinary N-acetyl-beta-D-glucosaminidase activity, five (12%) increased excretion of beta 2-glycoprotein-1, and five (12%) increased loss of retinol binding protein. Among 47 patients with an albumin excretion of 0.9-7.9 mg/mmol creatinine, 42 (89%) had increased urinary N-acetyl-beta-D-glucosaminidase, 23 (49%) an increased output of beta 2-glycoprotein-1, and 16 (34%) a raised excretion of retinol binding protein. The excretion of these markers of tubular defects seldom exceeded two and a half times the upper reference limit and the differences between the findings in the insulin dependent and non-insulin dependent patients with similar albumin excretion were small and insignificant. CONCLUSIONS In diabetic patients with a negative dipstick test for proteinuria: (a) assay of urinary beta 2-glycoprotein-1 may be a more sensitive test for the detection of impaired tubular reabsorption of protein than measurement of retinol-binding protein; (b) assay of N-acetyl-beta-D-glucosaminidase can detect tubular injury at a time when protein reabsorption remains normal; and (c) impaired renal tubular function may be present in the absence of evidence of glomerular malfunction.