Aims. This study investigated whether impulsivity measured in 12-14-year-olds could predict problem gambling in late adolescence, above and beyond other personality factors such as aggressiveness and anxiety. Design. A prospective-longitudinal design was used, thus overcoming limitations of past studies which used concurrent or retrospective designs. Participants and measurements. The sample included 154 boys living in economically deprived neighborhoods. Impulsivity measures comprised self-reports, teacher ratings and laboratory tasks, and were administered during early adolescence. Gambling behavior was assessed at age 17 using a self-report measure. Early gambling behavior and socio-demographic information were also collected for control purposes. Findings. Results revealed that a self-report measure of impulsiveness and a card-sorting task significantly predicted problem gambling, even after controlling for socio-demographic variables, early gambling behavior and other personality variables such as aggressiveness and anxiety. Moreover, the predictive link held across all levels of aggressiveness and anxiety. Both impulsivity measures seemed to tap an inability to foresee negative consequences and an inability to stop responding despite unfavorable contingencies. Conclusion. These findings suggest that disinhibited individuals with response modulation deficits are at risk for problem gambling, thus supporting the DSM-IV classification of pathological gambling as an impulse control deficit.