Objective: Determine factors that affect responsiveness to cognitive rehabilitation (CR) interventions in service members (SMs) who sustained mild traumatic brain injury (mTBI). Method: 126 SMs with a history of mTBI 3 to 24 months postinjury participated in a randomized clinical trial of one of four, 6-week treatment arms: (a) psychoeducation, (b) computer-based CR, (c) therapist-directed manualized CR, and (d) therapist-directed CR integrated with cognitive–behavioral psychotherapy. Practice-adjusted reliable change scores (RCS) were calculated for the three primary outcome measures: Paced Auditory Serial Addition Test (PASAT), Symptom Checklist-90 Revised (SCL-90–R) Global Severity Index (GSI), and Key Behaviors Change Inventory (KBCI). Hierarchical logistic regression was used to predict RCS. Variables considered were: (a) demographic, (b) injury characteristics, (c) comorbid mental health conditions, (d) nonspecific treatment variables (i.e., team vs. no-team milieu), and (e) specific treatment elements. Results: No predictor variables were associated with RCS improvements on the PASAT or the SCL-90–R. Comorbid depression (p < .02) and team-treatment milieu (p < .02) were associated with RCS improvement on the KBCI. Specific CR (ps > .65) and psychotherapy treatments (p > .26) were not associated with improvements on any outcome. There was evidence that self-administered computer CR was not only not beneficial, but negatively associated with cognitive and neurobehavioral improvement. Conclusions: Although reliable improvements were found on the PASAT and KBCI, no specific treatment intervention effects were found. Rather, comorbid depression and team-milieu treatment environment were associated with improvement, but only on the KBCI. Comorbid depression was associated with higher rates of improvement.