This study compared the characteristics of 4,117 women treated in publicly funded residential drug treatment programs in Los Angeles County between 1987 and 1994 by pregnancy status and program gender composition, that is, women-only and mixed-gender programs. A logistic regression analysis determined the predictors of program completion. Women in women-only programs were more likely than women in mixed-gender programs to be pregnant, homeless, or on probation; to use methamphetamines; to use alcohol; and have prior drug treatment. Pregnant women were younger, more likely to be homeless, had fewer years of drug use, were more often referred by other service providers, and were less likely to have injected drugs or have prior drug treatment than non-pregnant women. Although women in women-only programs had more problems, they spent more time in treatment and were more than twice as likely to complete treatment as compared with women in mixed-gender programs.