A core prediction of the reformulated model of learned helplessness and depression (Abramson, Seligman, & Teasdale, 1978) is that when confronted with the same negative life event, people who display a generalized tendency to attribute negative outcomes to internal, stable, or global factors should be more likely to experience a depressive mood reaction than people who typically attribute negative outcomes to external, unstable, or specific factors. We tested this prediction with a prospective design in a naturalistic setting by determining whether the content of college students' attributional styles at one point in time predicted the severity of their depressive mood response to receiving a low grade on a midterm exam at a subsequent point in time. Consistent with the prediction, students with an internal or global attributional style for negative outcomes at Time 1 experienced a depressive mood response when confronted with a subsequent low midterm grade, whereas students with an external or specific attributional style for negative outcomes were invulnerable to this depressive mood response. In contrast to the results for the internality and globality dimensions, students' scores along the stability attribution dimension were not correlated with the severity of their depressive mood response to the low midterm grade. In the absence of a negative life event (i.e., receipt of a high midterm grade), students' generalized tendencies to make internal or global attributions for negative outcomes at Time 1 were not significantly correlated with their subsequent changes in depressive mood although there was a nonsignificant positive correlation between severity of depressive mood response and the tendency to make global attributions for negative outcomes at Time 1.