Blood lead concentrations in children have decreased dramatically in the last 20 years pari passu with the decrease in the amount of lead used in gasoline nationwide. The NHANES study of 1988-91 reported that the incidence of lead concentrations ≥ 25 μg/dL had fallen to 0.5% nationally. Blood Pb values in Northeast continued to be greater than in the rest of the country(33% greater than in the Midwest and over 200% those in the south and west of the US). However, the level in Massachusetts was similar to the national average. To explain the lower blood Pb in MA than in neighboring states, we compared average measures of childhood lead exposure in census tracts from Worcester County, MA, and Providence County, RI. These counties are geographically close, similar in metropolitan size, demographics, climate, and air lead concentrations. Pre-1950's houses comprise 45% of Worcester and 50% of Providence county housing stock. However, MA has aggressively enforced legal requirements for abatement of housing lead paint hazards since 1973. RI has had no legal abatement requirements before 1992. Both states have laws which require universal screening of young children. In a small areas analysis, we calculated mean blood lead in each of 154 census tracts (CTs) from Worcester County, MA (9787 children, 0-4 years, screened in 1993), and 136 Providence County, RI CTs (17982 children, 0-4 years, screened in 1994). We hypothesized that mean blood lead would be higher among RI CTs even after controlling for differences in US 1990 census-derived demographics and housing characteristics because of the cumulative effects of MA housing policy. Blood lead was ≥ 25 μg/dL in 0.5% of MA children and 5.2% of RI children. Mean blood lead was 5.6 μg/dL (std dev = 1.3) in the average MA CT, whereas it was 8.3 μg/dL (std dev = 2.4) in the average RI CT, an unadjusted difference 2.7 μg/dL. After adjusting for 12 sociodemographic and housing variables with multiple regression analysis, mean blood Pb in children in the average RI CT remained 33% higher than in the average MA CT (1.8 μg/dL, p<.0001). The MA-RI difference is completely explained by a significantly higher slope coefficient for the predicted relationship between pre-1950's housing in RI (β=0.050 in old RI houses) vs MA (β=0.014 in old MA houses), suggesting that this adjusted RI-MA difference is due to factors associated with old housing, of which the leading candidate factor is the cumulative effects of 23 years of enforced abatement of lead hazards in MA homes. Confirmation of this finding would have major policy implications. However, direct measurement of housing variables which model abatement will be necessary to confirm this premise.