<p>Resting-state functional connectivity is increasingly used to measure neural engagement following transcranial magnetic stimulation (TMS), but whether connectivity changes are specific to the stimulation target and clinically meaningful remains unclear. In a head-to-head trial, thirty-six patients with depression and anxiety were randomized to 30 TMS sessions at one of two circuit-based targets: a dorsolateral prefrontal target hypothesized to improve “dysphoric“ symptoms, and a dorsomedial prefrontal target hypothesized to improve “anxiosomatic” symptoms. We tested whether: (1) TMS selectively engaged the targeted circuit; (2) baseline TMS site connectivity to the targeted circuit predicted circuit engagement; (3) either metric correlated with improvement on the Beck Depression and Beck Anxiety Inventories. Twenty-nine participants with pre- and post-treatment MRI were included. Target engagement was defined as change in average connectivity within each circuit and compared between target groups; associations were measured using Spearman’s partial correlation, controlling for relevant covariates. While connectivity decreased within both circuits, the effects were not clearly target-specific and did not predict symptom improvement. In contrast, stronger baseline TMS site connectivity to the targeted circuit predicted greater connectivity change (<i>r</i> = −0.61, <i>p</i> &lt; 0.001) and anxiety improvement (<i>r</i>_<i>w</i> = 0.47, <i>p</i> = 0.004). These effects were driven by the anxiosomatic target group. Negative findings in the dysphoric target group may reflect greater inter-individual variability in dysphoric circuit topography, as suggested by post-hoc analyses. Within the constraints of a modest sample size, results suggest that baseline TMS site connectivity may represent a more informative biomarker than TMS-induced connectivity changes, and that circuit architecture should be considered when defining targeting protocols.</p>