Transcranial magnetic stimulation (TMS) is effective for major depressive disorder (MDD) despite imprecise scalp-based targeting. Retrospective analyses suggest that targeting one brain circuit improves “dysphoric” symptoms, while targeting a different brain circuit improves “anxiosomatic” symptoms. Here, we tested this hypothesis prospectively. Individuals with moderate-to-severe MDD and moderate-to-severe anxiety (n=40) were randomized to receive TMS with dysphoric circuit targeting (dorsolateral prefrontal cortex) or anxiosomatic circuit targeting (dorsomedial prefrontal cortex). As hypothesized, dysphoric circuit targeting (n=16) improved BDI more than BAI (ratio 1.08, IQR 0.69-2.02), while anxiosomatic circuit targeting (n=20) improved BAI more than BDI (ratio 0.70, IQR 0.01-1.01) (Wilcoxon rank-sum test p=0.0195). This result was driven by larger improvements in anxiety with -anxiosomatic circuit targeting (p=0.0301). Thus, TMS targeting different brain circuits differentially modulates comorbid anxiety and depression symptoms. Symptom- and circuit-specific trial designs may improve power and better control for placebo and non-specific effects.