The relation between alcohol outcome expectancies (EXP) and alcohol use was prospectively examined over 3 years in a mixed-gender sample of college students (N = 465) at low and high risk for the development of alcoholism. Alcohol use remained fairly stable over 4 years, but EXP decreased significantly over the course of the study. Structural equation modeling techniques were used to examine reciprocal relations between EXP and alcohol use over 1- and 3-year intervals. Reciprocal prospective effects were demonstrated, but the nature of these effects appears dependent on the interval between measurement periods. Conceptually, these findings indicate both an etiologic role for EXP in predicting future alcohol use, and the influence of alcohol consumption on the development and maintenance of EXP. Methodologically, they point to the importance of the consideration of measurement interval in longitudinal research.