Background: Optimising treatments for patients with treatment-resistant depression (TRD) is key to reducing the burden of this severe illness. The anti-glucocorticoid medication metyrapone has mixed evidence supporting a role as a possible augmentation treatment in TRD. The degree of treatment resistance in depression has been associated prospectively and retrospectively with elevated inflammation, and inflammatory activity may influence responses to antidepressant treatments. Aims: To investigate whether levels of pro-inflammatory cytokines are associated with clinical outcomes to metyrapone or placebo. Methods: A double-blind RCT randomised patients with TRD to 3 weeks of placebo or metyrapone augmentation to ongoing serotonergic antidepressants. No benefit of metyrapone was reported in the primary analysis. The current study assessed levels of pro-inflammatory proteins interleukin-6 (IL-6), tumour necrosis factor (TNFα), c-reactive protein (CRP) and interleukin-10 (IL-10) before randomisation and after treatment as potential moderators and/or mediators of clinical outcomes. Results: The three pro-inflammatory proteins (but not IL-10) were elevated in this sample of patients with TRD compared to a non-affected control group. High pre-treatment IL-6 levels predicted a poorer response in the trial overall but did not moderate response to metyrapone versus placebo. Changes in IL-6 indirectly mediated depression outcome, with metyrapone increasing IL-6 levels and IL-6 increase associated with a poorer outcome on depression. Other inflammatory proteins did not mediate or moderate treatment outcomes. Interpretation: Metyrapone is hypothesised to have a therapeutic effect in depression on the basis of inhibiting the synthesis of cortisol. In this study, metyrapone did not reduce cortisol, possibly due to glucocorticoid system overcompensation). The mediation effect of IL-6 may support this and perhaps help to indicate why the treatment was not effective.