Volumetric analysis of urinary red blood cells (RBCs) was performed in two groups of patients: (a) 91 with glomerulonephritis (GN), 158 with non-glomer-ular diseases (non-GN) and 53 controls without haematuria; (b) 97 with GN and 41 with non-GN diagnoses and ≧ 2+ haematuria, analysed after modified sample preparation (isotonic dilution and haemolysis to eliminate non-RBC debris). In group A, diagnostic ranges were established for modal, mean and differential (urinary-peripheral blood modal) RBC volumes to differentiate GN from non-GN sources of blood loss according to RBC size. 54% (135/249) of modal volumes and 68% (124/183) with ≧2+ haematuria were within a diagnostic range of values, i.e. 40-180 fl (sensitivity ), and the source of haematuria was correctly identified in 85% (115/135) and 87% (108/124) of these, respectively (specificity). Of the remainder, with modal volumes 180 fl, 45% (51/114) were from GN and 55% (63/114) from non-GN specimens, along with 94% (48/53) controls without haematuria (non-diagnostic analyses). In contrast, whilst 90% (74/82) of mean volumes ≧ 110 fl and 94% (62/66) of differential volumes ≧ 0 fl were from non-GN specimens, values below these were common to both diagnostic groups so that diagnostic sensitivities were reduced to 38% (82/219) and 34% (66/193), respectively. In group B, haemolysis enabled extension of the diagnostic range of GN modal volumes down to 30 fl, whilst specificity of the test was enhanced with the modified sample preparation to 100% (86/86) versus 87% for those with ≧ 2+ haematuria from group A (p < 0.001). These findings illustrate the value of a standardized methodology for the assessment of urinary RBC volumes in the investigation of patients with unexplained urinary blood loss: diagnostic accuracy is enhanced with a classification of modal volumes and modified sample preparation in specimens with ≧ 2+ dipstick