Objectives. To determine if directly observed treatment with once-weekly fluoxetine would reduce depression symptom severity and improve HIV outcomes among homeless and marginally housed HIV+ adults. Methods. This was a non-blinded, randomized controlled trial of 137 HIV+ homeless and marginally housed persons who met DSM-IV criteria for Major Depressive Disorder, Dysthymia, or Minor Depressive Disorder. The intervention was directly-observed once-weekly fluoxetine for 24 weeks, then self-administered once-weekly fluoxetine for 12 weeks. The Hamilton Depression Rating Scale was the primary outcome; response was defined as a 50% reduction from baseline, and remission was defined as a score below 8. Secondary outcome measures were: Beck Depression Inventory-II, antiretroviral uptake, antiretroviral adherence as measured by unannounced pill count, and HIV-1 RNA viral suppression to <50 copies/mL. Results. The intervention reduced depression symptom severity (b= 1.97; 95% CI, -0.85 to -3.08; P<0.001) and increased response (AOR=2.40; 95% CI, 1.86-3.10; P<0.001) and remission (AOR=2.97; 95% CI, 1.29-3.87; P<0.001). Similar results were obtained with the Beck Depression Inventory-II. No statistically significant differences in secondary HIV outcomes were observed. Conclusions. Weekly fluoxetine treatment of depression among HIV+ homeless and marginally housed adults resulted in reduced depression symptom severity, but not improved HIV outcomes, over nine months of follow-up.