Background: Behavioral activation is an effective treatment for depression that is theorized to facilitate structured increases in enjoyable activities that increase opportunities for contact with positive reinforcement; to date, however, only few mechanistic studies focused on a standalone intervention. Method: Interventions using internet-based behavioral activation or psychoeducation were compared based on data from a randomized-controlled trial of 313 patients with major depressive disorder. Activation level and depression were measured fortnightly (baseline, Weeks 2, 4, 6, 8, 10), using the Patient Health Questionnaire-9 and the Behavioral Activation for Depression Scale-Short Form, respectively. Analysis was performed to determine if a change in activation level mediated treatment efficacy. Results: Latent growth modeling showed that internet-based behavioral activation treatment significantly reduced depressive symptoms from baseline to the end of treatment (standardized coefficient = −.13, p = .017) by increasing the rate of growth in the activation level (mediated effect estimate = −.17, 95% CI [−.27, −.07]. Results from mixed effects and simplex models showed that it took 4 weeks before mediation occurred (i.e., a significant change in activation that led to a reduction in depressive symptoms). Conclusion: Activation level likely mediated the therapeutic effect of behavioral activation on depression in our intervention. This finding may be of significant value to clinicians and depressed individuals who should anticipate a 4-week window before seeing a prominent change in activation level and a 6-week window before depressive symptomatology reduces. Future research must consolidate our findings on how behavioral activation works and when mediation occurs.