BACKGROUND: This study is a confirmatory efficacy trial of two treatments for winter seasonal affective disorder (SAD): SAD-tailored group cognitive-behavioral therapy (CBT-SAD) and light therapy (LT). In our previous efficacy trial, post-treatment outcomes for CBT-SAD and LT were very similar, but CBT-SAD was associated with fewer depression recurrences two winters later than LT (27.3% in CBT-SAD vs. 45.6% in LT). CBT-SAD engaged and altered a specific mechanism of action, seasonal beliefs, which mediated CBT-SAD's acute antidepressant effects and CBT-SAD's enduring benefit over LT. Seasonal beliefs are theoretically distinct from LT's assumed target and mechanism: correction of circadian phase. This study applies the experimental therapeutics approach to determine how each treatment works when it is effective and to identify the best candidates for each. Biomarkers of LT's target and effect include circadian phase angle difference and the post-illumination pupil response. Biomarkers of CBT-SAD's target and effect include decreased pupillary and sustained frontal gamma-band EEG responses to seasonal words, which are hypothesized as biomarkers of seasonal beliefs, reflecting less engagement with seasonal stimuli following CBT-SAD. In addition to determining change mechanisms, this study tests the efficacy of a "switch" decision rule upon recurrence to inform clinical decision-making in practice. METHODS: Adults with SAD (target N = 160) will be randomzied to 6-weeks of CBT-SAD or LT in winter 1; followed in winter 2; and, if a depression recurrence occurs, offered cross-over into the alternate treatment (i.e., switch from LT➔CBT-SAD or CBT-SAD➔LT). All subjects will be followed in winter 3. Biomarker assessments occur at pre-, mid-, and post-treatment in winter 1, at winter 2 follow-up (and again at mid-/post-treatment for those crossed-over), and at winter 3 follow-up. Primary efficacy analyses will test superiority of CBT-SAD over LT on depression recurrence status (the primary outcome). Mediation analyses will use parallel process latent growth curve modeling. DISCUSSION: Consistent with the National Institute of Mental Health's priorities for demonstrating target engagement at the level of Research Domain Criteria-relevant biomarkers, this work aims to confirm the targets and mechanisms of LT and CBT-SAD to maximize the impact of future dissemination efforts. TRIAL REGISTRATION: ClinicalTrials.gov identifier: NCT03691792 . Registered on October 2, 2018.