OBJECTIVE: To compare blood lead (BPb) poisoning screening strategies in light of the 1997 recommendations by the Centers for Disease Control and Prevention, Atlanta, Ga. DESIGN: Cost-effectiveness analysis from the perspective of the health care system to compare the following 4 screening strategies: (1) universal screening of venous BPb levels; (2) universal screening of capillary BPb levels; (3) targeted screening of venous BPb levels for those at risk; and (4) targeted screening of capillary BPb levels for those at risk. Costs of follow-up testing and treatment were included in the model. RESULTS: Only universal venous screening detected all BPb levels of at least 0.48 micromol/L (10 microg/dL). Universal capillary screening detected between 93.2% and 95.5% of cases, depending on the prevalence of elevated BPb levels. Targeted screening was the least sensitive strategy for detecting cases. Venous testing identified between 77.3% and 77.9% of cases, and capillary testing detected between 72.7% and 72.8% of cases. In high-prevalence populations, universal venous screening minimized the cost per case ($490). In low- and medium-prevalence populations, targeted screening using venous testing minimized the cost per case ($729 and $556, respectively). In all populations, regardless of screening strategy, venous testing resulted in a lower cost per case than capillary testing. Sensitivity analyses of all parameters in this model demonstrated that this conclusion is robust. CONCLUSIONS: Universal screening detects all cases of lead poisoning and is the most cost-effective strategy in high-prevalence populations. In populations with lower prevalence, the cost per case detected using targeted screening is less than that of universal screening. The benefit of detecting a greater number of cases using universal screening must be weighed against the extra cost of screening. Regardless of whether a strategy of universal or targeted screening is used, the cost per case using venous testing is less than that of capillary testing.