The last decade has witnessed an increase in the number of moderate to large-scale nonpharmacologic stroke recovery trials. While a majority, having tested the superiority of a particular evidence-based intervention, returned negative findings, the rehabilitation research community has gained an important perspective for future efforts. We offer our interpretation first, on why most of the past decade’s trials failed in the sense of not supporting the primary superiority hypothesis, and, second, we provide our perspective on how to solve this problem and thereby inform the next generation of neurorehabilitation clinical trials. The first large-scale randomized controlled trial (RCT) ever conducted in neurorehabilitation was the Extremity Constraint Induced Movement Therapy Evaluation (EXCITE) trial. The majority of stroke recovery trials that followed were based on a prevailing, but as yet immature science of brain-behavior mechanisms for recovery and limited practical know-how about how to select the most meaningful outcomes. The research community had been seduced by a set of preclinical studies, ignited by the 1990’s revolution in neuroscience and an oversimplified premise that high doses of task-oriented training was the most important ingredient to foster recovery. Here, we highlight recent qualitative and quantitative evidence, both mechanistic and theory-driven, that integrates crucial social and personal factors to inform a more mature science better suited for the next generation of recovery-supportive rehabilitation clinical trials.