The sample included 152 males and 151 females. Information was collected by means of a 59-item life stress measure that included 16 controllable life events and 34 uncontrollable life events, a 24-item Abusable Substance Locus of Control Scale, and a survey regarding drug use. Information was gathered from 303 students in the 1989-90 school year and from 178 of these students in the 1991-92 school year. Results revealed that the students lost to followup were similar to those measured at both time periods with respect to socioeconomic status, gender, drug use at the initial measurement, and controllable and uncontrollable stressors. Findings also revealed that controllable life stress contributed to six regression equations predicting children's drug use both cross-sectionally and longitudinally, even when selected demographic and personal variables were controlled. Data supporting a similar role for uncontrollable life stress were not found. Findings suggest that a child with a high number of controllable life events may be at increased risk for drug use. Further research should focus on the role life stress has in juvenile drug use. Tables and 35 references (Author abstract modified)