The proportion of dysmorphic red cells (DRC) in the urinary sediment and their mean corpuscular volume (MCV) have been claimed to discriminate between glomerular and postglomerular sources of haematuria. To determine the diagnostic value of urinary DRC and MCV, we searched the literature and critically reviewed 21 published studies using a predetermined set of criteria for evaluation. All studies originated from referral centres. Interobserver variability in identifying urinary DRC was reported in four studies and found to be unacceptably large in one. Although reproducible over different samples of the same individual, urinary MCV was unreliable in cases of low-grade haematuria because of interfering debris. Weighted averages and 95% confidence limits of the sensitivity and specificity of the DRC proportion for glomerular disease were 0.88 (0.86-0.90) and 0.95 (0.93-0.97), respectively; those of a low MCV were 1.00 (0.98-1.00) for sensitivity and 0.87 (0.80-0.91) for specificity. Sensitivity and specificity values derived from in-patients were slightly higher than those in referred outpatients. No studies of urinary DRC or MCV in patients with incidentally detected microhaematuria in the primary care setting were found. We conclude that at present the diagnostic value of urinary DRC and MCV is limited. In referral centres, that is, in patients with a high probability of postglomerular haematuria, the test cannot rule out urological lesions, because its specificity for glomerular disease may be as low as 0.80. In the primary care setting, that is, in unselected patients with incidentally detected low-grade haematuria, the accuracy of the test has not been studied but may be even lower. The use of urinary DRC or MCV as an indicator of the source of haematuria is in need of further experimental development and confirmation.