This article considers the validity of estimates of intentional violence using a statewide injury recording system: the Alaska Trauma Registry (ATR). One benefit of using data from an injury surveillance system is that its records are generated without police involvement, thereby reducing the likelihood of undercounting. However, there is reason to suspect that measures of violence derived from injury surveillance systems do not accurately measure underlying levels of assault in a population and are instead partly a reflection of influential patient characteristics. The ATR was used to compare assault injury cases of Alaska Natives with those of non-Natives to determine if patient characteristics varied between racial groups and to determine if those characteristics served to bias comparisons of injury rates. Results indicate that differences in assault injury rates are partly attributable to a greater proportion of Alaska Native patients being hospitalized for injuries of only minor or moderate severity. (Contains 4 tables and 2 figures.)