Using a four-way asymmetrical analysis, logit-models were tested to examine the relationships between the response variable, the types of crimes charged (nonaggressive versus aggressive), and a set of exploratory variables, ethnicity, drug test results, and frequency of alcohol use. Results indicated a complex but interpretable relationship between drug use, alcohol use patterns, and aggressive crimes. A surprising findings was that more aggressive crimes were committed by all participants testing negative for drugs. Mexican-Americans with frequent alcohol use and testing positive for drugs were twice as likely to commit a violent crime as were whites in the same subgroup. However, the white frequent drinker who does not use other drugs was the most violent group. Findings indicated that public perceptions linking violence with Mexican-Americans obscure the true situation and that prevention efforts should be targeted not only to drug use in the Mexican- American community but also toward whites' belligerent drinking. Tables and 47 references (Author abstract modified)