In a previous study, we showed that in our region the prevalence of lead poisoning (Pb > or = 10 micrograms/dL [0.483 mumol/L]) was high (29%) among urban infants. The objectives of this study were to assess the prevalence of lead poisoning over time in this cohort to determine the best screening strategy. Eighty-nine percent (115/129) of infants who were initially screened for lead poisoning between 6 and 12 months of age had a repeat venous lead level by age 3 years (mean = 19.8 months). The majority (70%) lived in a major city and had private medical insurance (80%). The initial and repeat mean venous lead levels were similar for the overall population (lead1 = 8.7 micrograms/dL +/- 6.2 [0.420 +/- 0.299 mumol/L] and lead2 = 7.7 micrograms/dL +/- 6.8 [0.371 +/- 0.328 mumol/L]). However, the percentage of lead-poisoned (Pb > or = 10 micrograms/dL or 0.483 mumol/L) children whose levels were > or = 15 micrograms/dL (0.724 mumol/L) increased from 28% to 48% on repeat. The utility of the initial screen in predicting a second elevated lead level > or = 20 micrograms/dL (0.966 mumol/L) was as follows: sensitivity = 33%, specificity = 97%, positive predictive value = 40%, negative predictive value = 96%. We conclude that in this population, a single screen at 18-24 months of age would reduce the need for multiple venipunctures and would detect those children in need of an intervention.