A path analysis test of the attributional model of depression was conducted in a naturalistic achievement setting (N = 77). Relatively depressed and nondepressed students reported their attributions on an academic test, and then their subsequent test performance were measured. In support of the attributional model of depression, the results indicated that for successful outcomes the attribution dimension of stability (i.e., stable vs. unstable across different situational contexts) served to buffer the influence of depression on later test performances. The results thus show that stability attributions for a successful behavior, defined in terms of consistency across different situational contexts, play a mediating role in the behavioral consequences of depression.