Although inpatient multidisciplinary pain management programs (PMPs) are effective for chronic pain, not all patients benefit equally and there is limited evidence regarding predictors of outcome. This meta-analysis aimed to identify patient or program characteristics associated with outcomes from inpatient PMPs, and to examine the time course of effects following discharge. Medline, EBSCO, and Scopus were searched to identify articles reporting outcomes from inpatient multidisciplinary PMPs. Information was extracted on study design, participant and program characteristics, and outcomes. Effect sizes were computed for pain, physical function, depression, anxiety, and mental health outcomes. Study-level predictors of outcome were investigated with moderator analyses and meta-regression. A risk of bias assessment and sensitivity analyses were conducted and the GRADE criteria for prognostic studies were applied to assess confidence in findings. In all, 85 studies (111 cohorts; 15,255 participants) were included. Three quarters of studies demonstrated low risk of bias. Larger effect sizes (for at least 1 outcome measure) occurred in studies where participants had more severe pain (greater intensity/longer duration), participants with alcohol or drug problems were not excluded, samples comprised mixed pain conditions, and programs included a cognitive component and/or a passive therapy component. Effect sizes for pain and physical function were maintained at follow-up, but effect sizes for depression and anxiety declined over time. Inpatient multidisciplinary PMPs may be well suited to patients with severe or long-lasting pain. Programs should adopt broad patient inclusion criteria, and outcomes were similar for programs based on cognitive-behavioral versus mindfulness/acceptance-based therapies.