Habitual caffeine consumption protects against depression but through unclear mechanisms. Although habitual caffeine use predicts cortisol release in response to other acute stressors (e.g., exercise), this is less examined with lab-based psychosocial stress in healthy adults. Furthermore, caffeine-induced cortisol increases may mask theory-predicted cortisol blunting to robust stress in people with elevated depression risk. In two samples, we tested whether acute (same-day) and habitual caffeine use would predict greater cortisol reactivity to lab-based stress, and whether caffeine would "mask" the effect of a depression risk factor, trait rumination, on blunted cortisol reactivity. In sample 1, N = 128 emerging adults completed one of three Trier Social Stress Test conditions: nonevaluative control, ambiguously evaluative intermediate, or explicit negative evaluative. In sample 2, N = 148 emerging adults completed either a control or negative evaluative condition. In both samples, multilevel growth curve modeling indicated that habitual caffeine use ( t = -1.99, p = .048; t = -2.73, p = .007, samples 1 and 2, respectively) but not acute caffeine use predicted heightened cortisol reactivity as a function of condition. In sample 1, the relationship between condition, rumination, and blunted cortisol was evident only in caffeine nonusers, which differed from users ( t = 2.82, p = .005), but in sample 2, the predicted blunting pattern was evident regardless of caffeine use. The results provide evidence that habitual caffeine use is associated with greater cortisol release under psychosocial lab-based stress and may mask the influence of psychosocial variables; future research should examine whether habitual caffeine-induced cortisol release has behaviorally activating effects that protect against depression.