BACKGROUND: Many patients with major depressive disorder have treatment-resistant depression, defined as no adequate response to two consecutive courses of antidepressants. Some evidence suggests that antiglucocorticoid augmentation of antidepressants might be efficacious in patients with major depressive disorder. We aimed to test the proof of concept of metyrapone for the augmentation of serotonergic antidepressants in the clinically relevant population of patients with treatment-resistant depression. METHODS: This double-blind, randomised, placebo-controlled trial recruited patients from seven UK National Health Service (NHS) Mental Health Trusts from three areas (northeast England, northwest England, and the Leeds and Bradford area). Eligible patients were aged 18-65 years with treatment-resistant depression (Hamilton Depression Rating Scale 17-item score of ≥18 and a Massachusetts General Hospital Treatment-Resistant Depression staging score of 2-10) and taking a single-agent or combination antidepressant treatment that included a serotonergic drug. Patients were randomly assigned (1:1) through a centralised web-based system to metyrapone (500 mg twice daily) or placebo, in addition to their existing antidepressant regimen, for 21 days. Permuted block randomisation was done with a block size of two or four, stratified by centre and primary or secondary care setting. The primary outcome was improvement in Montgomery-Åsberg Depression Rating Scale (MADRS) score 5 weeks after randomisation, analysed in the modified intention-to-treat population of all randomly assigned patients that completed the MADRS assessment at week 5. The study has an International Standard Randomised Controlled Trial Number (ISRCTN45338259) and is registered with the EU Clinical Trial register, number 2009-015165-31. FINDINGS: Between Feb 8, 2011, and Dec 10, 2012, 165 patients were recruited and randomly assigned (83 to metyrapone and 82 to placebo), with 143 (87%) completing the primary outcome assessment (69 [83%] in the metyrapone and 74 [90%] in the placebo group). At 5 weeks, MADRS score did not significantly differ between groups (21·7 points [95% CI 19·2-24·4] in the metyrapone group vs 22·6 points [20·1-24·8] in the placebo group; adjusted mean difference of -0·51 points [95% CI -3·48 to 2·46]; p=0·74). 12 serious adverse events were reported in four (5%) of 83 patients in the metyrapone group and six (7%) of 82 patients in the placebo group, none of which were related to study treatment. 134 adverse events occurred in 58 (70%) patients in the metyrapone group compared with 95 events in 45 (55%) patients in the placebo group, of which 11 (8%) events in the metyrapone group and four (4%) in the placebo group were judged by principle investigators at the time of occurrence to be probably related to the study drug. INTERPRETATION: Metyrapone augmentation of antidepressants is not efficacious in a broadly representative population of patients with treatment-resistant depression within the NHS and therefore is not an option for patients with treatment-resistant depression in routine clinical practice at this time. Further research is needed to clarify if such augmentation might benefit subpopulations with demonstrable hypothalamic-pituitary-adrenal axis abnormalities. FUNDING: Efficacy and Mechanism Evaluation (EME) programme, a UK Medical Research Council and National Institute for Health Research partnership.