The dependent variable was "risky decisions," which was defined broadly as either actual risky behaviors, such as driving while intoxicated, or as not making efforts to avoid risk, such as drinking at a party before making arrangements to get home safely. Independent variables were assertiveness, alcohol-related behaviors, the amount of alcohol consumed, the acceptability of drinking and driving, and previous drinking and driving/riding experience. Analyses show that alcohol-related behaviors, attitudes toward the acceptability of drinking and driving, and previous drinking and driving/riding experiences were all significant predictors of decisions about driving or riding while intoxicated. Indeed, the overall model accounted for 46 percent of risky drinking and driving/riding decisions. As predicted, older respondents had more previous experiences with driving while intoxicated and riding with intoxicated drivers than did younger respondents, and they reported that drinking and driving was more acceptable among their peers; however, contrary to expectations, there were no differences in the number of risky decisions made by the two age groups or between males and females. This study thus suggests that a personal history of drinking and driving/riding may be one of the major sources of information used by an older adolescent who is confronted with new decisions about combining alcohol with driving or riding. Prevention programs may be more effective if they address adolescents' experiences and the cognitive biases in thinking that can lead people into risky situations. 4 tables, appended vignettes, and 26 references