provide accurate symptom estimates than court-referred samples. A population-based sample large enough to capture a sufficient number of DUI offenders for analysis would provide the least biased estimates. The National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) is one large study that has good data on alcohol and drug use and psychiatric disorders. Surprisingly, although it contains questions about drinking and driving (Chou et al., 2006), it does not include an item about DUI arrest. An alternate sampling strategy is to use data from samples at high risk for alcohol dependence, which would likely have elevated rates of DUI relative to population-based samples, but which are not recruited from mandated treatment settings. The Collaborative Study on the Genetics of Alcoholism (COGA) is one such high-risk study. We used COGA data, in which 29% of men and 8% of women reported at least one DUI arrest, to examine gender differences in alcohol criteria endorsement, lifetime prevalence of non-substance psychiatric disorders, and illicit drug use and dependence for individuals who reported none, one, two, or three or more DUIs. We anticipated that examining gradations of disorder within a combination of two high-risk indicators, DUI and family vulnerability, might provide useful information
