The study found that the predictability of risky behavior for health in adolescence (smoking, sexual activity, poor diet, and suicidal tendencies) through self-reports compared with parental reports of psychopathology depended on the special type of health-risk behavior and the child's gender. Based on both self-reports and parental reports, the predictions of smoking and dietary behavior in the eighth grade from early indicators of psychopathology in the fourth grade were poor. Exceptions were the predictions of sexual activity and suicidal tendencies in boys, which were predicted from parental reports of acting-out behaviors and self-reported emotional states in grade four. The findings suggest it is useful to rely on information from different informants on problem behaviors in late childhood in order to predict adolescents' health-risk behaviors; however, the predictability of specific health-risk behaviors from self-reports and parental reports will vary, with gender being a factor. Just over half (55.5 percent) of the sample of 366 fourth-graders were girls. The data analyses were based on data for 245 children (55.1 percent girls) with complete self-report data for the assessments at grades four, six, and eight and complete parental reports at grades four and six. In grades four and six, the children completed the Youth Self Report, and their parents completed the Child Behavior Checklist. In grade eight, the children reported their health-risk behavior (smoking, sexual activity, dietary behavior, and suicidal tendencies). 6 tables, 1 figure, and 51 references