The current study used double-blind, placebo-controlled design to examine the effect of intranasal oxytocin (OT) on emotion recognition (ER) and visual attention in 60 outpatients presenting for assessment and treatment of emotional disorders. Our primary hypothesis was that OT would improve recognition of happy faces in depressed participants. The main effect of OT on ER accuracy, speed, and proportion of fixations in the eye region was not significant. Diagnostic group (i.e., presence/absence of a depressive disorder) moderated the effect of OT on ER, but not as expected: OT significantly slowed ER speed for all emotions in participants with anxiety disorders, but did not affect performance in participants with depressive disorders. Depressed participants fixated significantly less in the eye region of sad faces than anxious participants. Before OT can be used to target ER biases, additional research is needed to explicate the differential impact of OT on ER speed in patients with anxiety versus mood disorders.