Objective: To characterize long-term post injury drinking patterns of persons with traumatic brain injury and to identify injury-related and sociodemographic factors related to post injury consumption. Design: A cross-sectional design (n = 322) was first used to characterize drinking patterns at three time intervals. A longitudinal design (n = 73) was then used to examine changes in consumption patterns. XL analyses and Wilcoxon matched-pairs signed-ranks test (t-tests) were conducted to investigate the relationship between patients' reported alcohol use and time post injury. Spearman rank-order correlations, χ2 analyses, and Fisher's exact probability tests were used to examine relationships among drinking classifications and injury-related and sociodemographic characteristics. Setting: Rehabilitation center outpatient clinics affiliated with the four brain injury Model Systems programs. Patients: Patients with moderate and severe brain injury who received acute medical care and inpatient rehabilitation. Inclusion criteria were: patient > 16 years old, participation in follow-up medical and psychologic evaluations, and willingness to provide information regarding alcohol consumption. Main Outcome Measures: Quantity Frequency Variability Index, Disability Rating Scale, and Functional Independence Measure scores. Results: At each follow-up interval, the proportion of persons classified as moderate or heavy drinkers was generally comparable to that in the general population. Abstention rates, however, were markedly higher than those within the general population. For the cross-sectional sample, χ2 analyses revealed no significant differences in the distribution of drinking classifications across follow-up intervals. For the longitudinal group however, Wilcoxon matched-pairs signed-ranks tests revealed substantial differences in drinking ratings for individuals over time. A pattern of increasing consumption was indicated. Correlational analyses revealed higher consumption levels for younger persons and those with higher admission blood alcohol levels. Furthermore, relationships were observed between injury severity anti consumption levels. Higher levels of disability were associated with lower consumption rates. Conclusions: These data suggest that alcohol use patterns are dynamic during the first 4 years after injury. Also, long-term monitoring of alcohol use patterns may be cost-effective, particularly for younger persons and those with a preinjury history of problem drinking.