This study examined the prevalence of drug use in a nationally representative sample of 1989 recipients and 6840 nonrecipients of four welfare programs. Data from the 1995 National Household Survey on Drug Abuse (NHSDA) were analyzed using the conditional form of multiple logistic regression with matching of respondents on neighborhood of residence. Weighted proportions and variances accounting for the complex sample design of the NHSDA survey were estimated using the Taylor series linearization method. The results indicate that drug use is 50% more common in households with welfare recipients than in nonwelfare households. Programs making welfare eligibility contingent on the recipient working toward a drug-free lifestyle are worth examining, although a vigilant eye must be kept on the potential unintended consequences.