Fresh, uncentrifuged urine specimens obtained from children attending a renal clinic were examined microscopically for leucocytes and bacteria, and the results were compared with quantitative bacterial cultures. Of the 51 specimens giving colony-counts larger than or equal to 10-5 per ml., 6 (12 per cent) showed no microscopical bacteriuria and 22 (43 per cent) contained smaller than 10 leucocytes per c.mm.; only 3 specimens (6 per cent) showed neither bacteria nor excess cells. Of the 186 uninfected specimens, bacteria were seen microscopically in 32 (17 per cent) and an excess of leucocytes in 27 (15 per cent); in only 9 (5 per cent) were both observed. Leucocyte-counting alone is therefore an unreliable technique, but microscopy for both bacteria and leucocytes affords a simple and rapid method of preselecting urine specimens for culture, with a high probability of predicting the correct result. While not suitable for population screening, it is ideal for use in hospital clinics and wards, as well as in general practice.