This prospective study investigated in-treatment relapse in a sample of perinatal substance abusers in intensive outpatient treatment. Sixty-four female first-time admissions to a perinatal treatment program completed comprehensive psychological and psychosocial assessment before beginning treatment. Relapse was detected by urine toxicology screening and self-report. A regression analysis resulted in variable reduction, then survival analysis identified the impact of in-treatment relapse and other predictors on treatment length. Fifty-five percent of the subjects were classified as relapsers. Two risk factors for and six protective factors from in-treatment relapse were identified. The survival curves for relapsers and nonrelapsers did not differ until covariates were considered. Subjects with more severe consequences of drug use and less social exposure to drug use during treatment tended not to relapse during treatment, perhaps in order to prevent deterioration such as loss of children or incarceration. Relapse alone did not result in fewer treatment days. Few addiction characteristics were related to either in-treatment relapse or length of treatment. Rather, personality and demographic variables were more salient in both the regression and survival models. Treatment staff may need to reconsider their views of the meaning of relapse and should develop enhanced engagement and retention strategies for women at greater risk of relapse.