Recovery of wildlife and their ecosystems may require enhancement of vulnerable populations through human intervention. As a southern sea otter (Enhydra lutris nereis) enhancement strategy, surrogate rearing and releasing of stranded orphans provides these individuals with another opportunity of life in the wild, promotes species-specific development of behaviors necessary for survival and reproduction after release, and may bolster targeted populations and resiliency of nearshore ecosystems. To further advance surrogacy as a viable recovery strategy for this threatened subspecies, we comprehensively describe our methods and use a machine learning algorithm to monitor and evaluate program success. Specifically, we model 20 years (2002−2021) of rehabilitation outcomes for 64 pups (i.e., success or failure reacclimating to the wild) based on 34 parameters representing three release sites and five treatment stages: stranding, stabilization, surrogacy, pre-release, and release. We find that surrogacy is a successful rehabilitation method (75%), robust to all measured factors through release preparation, including idiosyncrasies in pup origin, age, early development, and pairings with female sea otters. Instead, key performance indicators of successful outcomes in the wild are environmental conditions experienced during release, such as favorable sea conditions and weather, localized dispersal, and minimal competition from local populations. Most importantly, success requires optimal values among all environmental conditions at chosen release sites because a decline in any one factor may result in failure. Our findings highlight the importance of continuously evaluating rehabilitation methods to enhance species recovery and may have broader application to restoring other threatened species and their ecosystems.