OBJECTIVE To assess the false positive rate of blood lead (BPb) determinations on samples obtained by fingerstick from children screened in an urban clinic. METHODS From a single fingerstick (N = 1573), blood was collected in a capillary tube for determining lead concentration (CPb) by graphite furnace and an additional sample was absorbed onto a filter paper for determining lead concentration (FPb) by atomic absorption spectrophotometry with Delves cup. Zinc protoporphyrin (ZPP) was measured immediately and a confirmatory venous lead (VPb) specimen was obtained at the same visit if the ZPP was > or = 35 micrograms/dL (0.6 mumol/L); children with either a CPb or FPb > or = 15 micrograms/dL (0.7 mumol/L) were later recalled for determining VPb. RESULTS For the 172 children who had a VPb on the same day as the screening tests, the false positive rates (95% confidence intervals) at a lead threshold of 15 micrograms/dL (0.7 mumol/L) were: CPb, 13.5% (6.7-20.3); FPb, 19.1% (11.8-26.4). Analyses using all 679 screens with a paired venous specimen (mean delay between screen and venous testing = 30 days) yielded much higher false positive rates (CPb, 31.3%; FPb, 46.0%). CONCLUSIONS Screening for lead poisoning is feasible within an urban pediatric clinic by direct measurement of lead concentration in blood samples obtained by fingerstick. The false positive rate that can be obtained is acceptable given the precision of measuring BPb concentration. Practitioners using a staged screening protocol may incorrectly attribute a higher false positive rate to the screening tests, when much of the error may be due to the temporal variability of BPb resulting from both biologic variability in BPb concentration and intermittent exposures.