We aimed to assess the utility of the percentage of G cells and dysmorphic erythrocytes in the diagnosis of glomerular hematuria in pediatric patients. We determined the percentage of dysmorphic erythrocytes and G1 cells in urine samples from patients with glomerulonephritis and other renal diseases of non-glomerular origin. There was excellent correlation and agreement between results obtained by counting the cells in counting chambers and in urine sediments. With cut-off values of > or =1%, > or =2%, > or =5%, and 10% for G1 cells, sensitivities for the detection of glomerulonephritis were 62%, 40%, 28%, and 10% and specificities were 89%, 95%, 95%, and 98%. For the dysmorphic erythrocytes cut-off values of > or =10%, > or =20%, 50%, and > or =90% gave respective sensitivities of 95%, 95%, 93%, and 62% and specificities of 24%, 34%, 43%, and 85%. In 38% of cases of biopsy-proven glomerulonephritis no G1 cells were found. For cut-off values of > or =50% dysmorphic erythrocytes and > or =1% G1 cells, the sensitivity and specificity were 60% and 91%. For cut-off values of > or =50% dysmorphic or > or =1% G1 cells, sensitivity and specificity were 93% and 44%. Our results show that neither the percentage of dysmorphic erythrocytes nor the G1 cell count is of adequate sensitivity and specificity to enable reliable differentiation of glomerular and non-glomerular hematurias. Both tests are needed to achieve >90% sensitivity and specificity.