This study explores the potential for the development of a surrogate for alcohol-involved traumatic injury. It presents a bivariate probit analysis that simultaneously models likelihoods of patients being tested for blood alcohol content (BAC) and having positive BACs given testing using 17,356 adult trauma cases selected from the California Regional Trauma Registry. It concludes that patient and injury characteristics predict both testing and BAC, and that a weighting scheme may be profitably used to determine changes in levels of alcohol-involved trauma in populations over time in the absence of empirical measurement of BAC.