Substance use recovery homes represent the largest residential, community-based post-treatment option for those with substance use disorders in the United States. It is still unclear what unique factors predict relapse after residents leave such homes. This study presents results of a longitudinal study of 497 residents who departed from 42 Oxford House recovery houses. We hypothesized that the predictors of post-departure relapse would be a multi-item measure of latent recovery, length of stay, and reason for departure from the home (voluntary vs. involuntary). Predictor effects were estimated as part of a two-step model with two outcomes: (a) lack of follow-up data after departure from the house, and (b) the likelihood of relapse. Determinants of missing follow-up data included less education, less time in residence, and involuntary departure. Relapse was more likely for individuals who were younger, had involuntarily left the house, and had lower values on the latent recovery factor. The implications of these important factors related to relapse following departure from residential recovery home settings are discussed.