Though it is widely believed that psychotherapy changes biology, this contention is largely based on observational data, subject to confounding. We report the first systematic review and meta-analysis of biological variables assessed, as outcomes or predictors of response, in randomized controlled trials of psychotherapy for adult depression. Fifty-one trials (5123 participants) and a pooled analysis were included. Biological markers were outcomes in 43 studies and predictors of treatment response in 9. At post-treatment, psychotherapy could not be distinguished from control conditions for glycaemic control (Hb1AC), 7 trials, Hedges' g= -.01, 95% CI -0.30 to 0.29, I 2 = 65% and cortisol concentration after-wake, 5 trials, Hedges' g= -.19, 95% CI -0.45 to 0.06, I 2 = 0%. Follow-up results were similar. For the other biological domains (immunological, neurobiological, inflammatory, weight, blood pressure), overall findings were mixed and often inconsistent. Few trials investigated prediction of response, with only neuroimaging markers showing promise. Across domains, we found limited evidence that benefits of psychological treatments for depression translate to biological outcomes., (Copyright © 2019 Elsevier Ltd. All rights reserved.)