Suicide is a major global health concern and a prominent cause of death in adolescents. Previous research on the prediction of suicide attempts has mainly focused on clinical or adult samples. To prevent suicides at an early stage, however, it is necessary to better understand the risk factors in a community sample of adolescents. In this study, we used elastic net regressions to predict suicide attempts of 17-year-olds in the Millennium Cohort Study, combining a large set of self- and other-reported variables from different categories (e.g., demographics, health, personality, emotions, sexuality, or victimization). Balanced predictive accuracy (= mean of sensitivity and specificity) was .84 when using data of the 17-year-olds’ assessment, but significantly lower (.76) when trying to predict suicide attempts three years in advance. We identified a set of essential variables that should be considered when screening for suicidal behavior and compare it to existing measures currently used.