Objective: The objective of the Genomics Used to Improve DEpression Decisions (GUIDED) trial was to evaluate the utility of pharmacogenomic testing to improve outcomes among patients with major depressive disorder (MDD) who had not responded to at least 1 prior medication trial. The objective of the present analysis was to assess outcomes for the subset of patients expected to benefit from combinatorial pharmacogenomic testing because they were taking medications with predicted gene-drug interactions. Methods: Participants (enrolled from April 14, 2014, to February 10, 2017) had an inadequate response to at least 1 psychotropic medication in the current episode of MDD. Patients were randomized to treatment as usual (TAU) or the guided-care arm, in which clinicians had access to a combinatorial pharmacogenomic test report to inform medication selection. Patients and raters were blinded to study arm through week 8. The following outcomes were assessed using the 17-item Hamilton Depression Rating Scale (HDRS-17): symptom improvement (percent change in HDRS-17 score), response (≥ 50% decrease in HDRS-17 score), and remission (HDRS-17 score ≤ 7). In the GUIDED trial, the primary endpoint of symptom improvement did not reach significance in the intent-to-treat cohort (P=.069). Here, a post hoc analysis of patients who were taking medications subject to gene-drug interactions at baseline as predicted by combinatorial pharmacogenomic testing (N = 912) is presented. Results: Among participants taking medications subject to gene-drug interactions at baseline, outcomes at week 8 were significantly improved for those in the guided-care arm compared to TAU (symptom improvement: 27.1% versus 22.1%, P=.029; response: 27.0% versus 19.0%, P=.008; remission: 18.2% versus 10.7%, P=.003). When patients who switched medications were assessed, all outcomes were significantly improved in the guided-care arm compared to TAU (P=.011 for symptom improvement, P=.011 for response, P=.008 for remission). Conclusions: By identifying and focusing on the patients with predicted gene-drug interactions, use of a combinatorial pharmacogenomic test significantly improved outcomes among patients with MDD who had at least 1 prior medication failure.