Mindfulness-based relapse prevention (MBRP) is an effective treatment for substance use disorders (SUD). However, evidence is primarily based on studies of closed groups, and few studies support flexible formats of MBRP, such as rolling groups. This nonrandomized, open trial evaluated feasibility, acceptability, dose-response relations, and mechanisms of rolling admission MBRP ("Rolling MBRP") offered as part of short-term residential treatment for SUD. Rolling MBRP was developed prior to the trial through an iterative process over several years. Participants included 109 adults (46% female, 74.3% racial/ethnic minorities, mean age=36.40). Rolling MBRP was offered to all patients in the program 2x/week and attendance was tracked. Outcomes were craving, self-efficacy, mental health, mindfulness, and self-compassion at discharge. Selfreported out-of-session mindfulness practice was examined as a mediator of attendance-outcome relations. Analyses involved multiple regression and mediation models. Feasibility was demonstrated by good attendance rates. Acceptability was demonstrated by high engagement in mindfulness practice and high satisfaction ratings. Total sessions attended did not predict outcomes at discharge. However, attending 2+ sessions (versus 1 or none) significantly predicted better mental health and higher mindfulness at discharge, and these effects were mediated by