Screening tests for bacteriuria based on two different principles were evaluated in1582 schoolgirls aged 5-11 years, and in 26 girls aged 3-16 years attending hospitalwith symptomatic urinary tract infection. Tests for hypoglucosuria, performed by a semi-automated fluorometric method and with Uriglox strips on early-morning urine samples voided after overnight fasting, gave unacceptably high false-negative rates (16.7% and 20.8% respectively). Oxoid and Uricult dipslides were immersed in fresh midstreamspecimens of urine obtained at school and read overnight incubation at 37 degrees C. Both gave comparable results, with low false-positive rates and no false-negative responses. The higher cost of screening by dipslides was halved by using the "dipstream" technique, which also gave no false-negative results. Its false-positive rate of 13.5% could be reduced to 1.8% by disregarding colony counts of 10-8 non-faecal organisms and over per litre, which appear unimportant in schoolchildren. Bacteriuria was found in 2.3% of the schoolgirls; 39% of them had symptons, compared with 7.2% of the healthy girls, and 25% showed vesicoureteric reflux, which in 17% was associated with renalscarring. Since the natural history of covert bacteriuria and its relationship withreflux and scarring remain undetermined further research is required. The dipstreamtechnique offers a simple, reliable, and comparatively cheap screening method which could also be applied in general practice.