The Centers for Disease Control and Prevention (CDC) has recommended using a five-item questionnaire at every regular office visit for all children 6 to 72 months of age to identify those at risk of high-dose exposure to lead. To determine how well the questionnaire identifies children with elevated lead levels. Comparison of results of the questionnaire, which is intended to identify children as being low-risk or high-risk for lead poisoning, with children's blood lead levels. A pediatric continuity clinic located in a major teaching hospital in Rochester, NY. A consecutive sample of 476 children aged 6 to 72 months without a prior history of lead poisoning who were seen in the clinic in July and August 1992, and (for those aged < 36 months) or 12 months (for those aged 36 to 72 months). Fingerstick lead samples were obtained from all children, and those > or = 15 micrograms/dL (0.72 mumol/L) were confirmed by subsequent venous lead determinations. Twenty-eight percent had blood lead levels > or = 10 micrograms/dL (0.48 mumol/L), 8% had levels > or = 15 micrograms/dL (0.72 mumol/L), and 5% had lead levels > or = 20 micrograms/dL (0.96 mumol/L). According to responses on the questionnaire, 44% were initially classified as low-risk, and 43% were high-risk. In 13% risk could not be determined because one or more items on the questionnaire had not been answered or were answered equivocally, whereas all other items were answered "No." Children for whom risk could not be determined were presumed to be at high risk and were added to that category, resulting in 56% of the study population so designated. The questionnaire was moderately effective in identifying children with elevated lead levels. Seventy percent of children having lead levels > or = 10 micrograms/dL (0.48 mumol/L) and 82% of children having lead levels > or = 15 micrograms/dL (0.72 mumol/L) had been classified as high-risk by the questionnaire. Children classified as low-risk were very unlikely to have elevated lead levels. Eighty-one percent of low-risk children had lead levels < 10 ug/dL, and 97% had lead levels < 15 ul/dL. An abbreviated questionnaire using only the first three items from the CDC questionnaire had almost identical effectiveness. In this clinical setting, in which children are largely urban, poor, and have a moderate to high risk of developing elevated lead levels, the CDC risk assessment questionnaire is effective in identifying children with elevated lead burdens. However, an abbreviated version of the questionnaire may be as effective as the complete questionnaire. Additional questions should be added to the questionnaire to improve its overall sensitivity, and the questionnaire should be tested in other settings to see if it is effective with children having different environmental exposures to lead.