Risk covariates of neuropsychological ability (NA) at treatment entry and neuropsychological recovery (NR) across 15 months were examined and replicated in 2 samples (Ns = 952 and 774) from Project MATCH, a multisite study of alcoholism treatments. NA at treatment entry was associated with age, education, and other covariates. Statistically significant mean increases in NA over time had small effect sizes, suggesting limited clinical significance of NR in the samples as a whole. However, initial NA and a combination of risk factors in direct and mediated pathways predicted a large proportion of individual differences in NR. Statistically significant but modest differential treatment effects on NR suggest that addiction treatments may need to be modified or developed to facilitate this important aspect of recovery.