The authors examined the impact on neuropsychological performance of past high lead exposure, followed by lower proximate lead exposure, in 2 groups of smelter workers selected on the basis of their patterns of blood lead levels (BLLs) over time. Prior to 1980 (past exposure), both groups had more than 90% of their BLLs > or = 40 microg/dl. During and subsequent to 1980 (proximate exposure), those subjects with more than 90% of their BLLs remaining at > or = 40 microg/dl were assigned to the high-high (H-H) pattern group (n = 40); whereas those with 90% of levels below 40 microg/dl were assigned to the high-low (H-L) pattern group (n = 40). Means (and standard deviations) for pre-1980 time-integrated blood lead (IBL) levels were similar for the H-H pattern [633.2 (202.2) microg/yr-dl] and the H-L pattern [556.5 (144.8) microg/yr x dl]; however, IBLs from 1980 on were significantly different [H-H pattern = 646.9 (58.70) microg/yr x dl and H-L pattern = 408.8 (46.37) microg/yr x dl; p < 0.0001]. Age, education, and years of employment were similar for both groups. Examination of 5 neuropsychological measures revealed that verbal memory was significantly better in the H-L pattern group than in the H-H group. Multivariate examination of the data showed that pattern of exposure contributed significantly to verbal memory performance, after adjustment for the covariates, current BLL, and IBL. A partial correlation analysis between verbal memory and IBL for past high exposure showed an association with H-H pattern, but none with H-L pattern. Pattern of BLLs over a working lifetime contributed unique variance to verbal memory. Absence of an association between past high lead exposure and verbal memory in the H-L pattern group suggests that reversibility of function may occur when proximate BLL is maintained below 40 microg/dl.