Abstract Objective: To examine the extent to which outcomes from traumatic brain injury differ as a function of time and can be predicted at discharge from inpatient rehabilitation. Design: Survey method employing cross-sectional analyses. Setting: An inpatient brain injury rehabilitation unit in a large midwestern academic medical center. Subjects: Ninety-five adults with traumatic brain injuries, 6 months to 5 years after inpatient rehabilitation, stratified by time postdischarge. Main Outcome Measures: Functional Independence Measure (FIM SM), Sickness Impact Profile (SIP), Medical Outcomes Survey SF-36, Community Integration Questionnaire (CIQ), Craig Handicap Assessment and Reporting Technique (CHART), Brief Symptom Inventory (BSI), Satisfaction With Life Scale (SWLS), and indices of current psychosocial functioning. Results: Substance abuse, need for supervision, life satisfaction, and selected subscales of the CIQ and CHART differed over the period 6 months to 5 years after discharge. Approximately 75% of the variance in current FIM scores, and 40% to 50% of CHART, CIQ, and SIP total scores, could be predicted at time of discharge. Conclusions: Outcomes over the first 5 years after discharge were dynamic, with most change being improvement, at least after the first 2 years. Important aspects of outcome could not be predicted based on premorbid characteristics, injury severity, and initial functional abilities.