Background Antidepressants are a first-line treatment for adults with moderate to severe major depression. However, many people prescribed antidepressants for depression don't respond fully to such medication, and little evidence is available to inform the most appropriate 'next step' treatment for such patients, who may be referred to as having treatment-resistant depression (TRD). National Institute for 1. Any psychological therapy versus antidepressant treatment alone, or another psychological therapy. 2. Any psychological therapy given in addition to antidepressant medication versus antidepressant treatment alone, or a psychological therapy alone. Primary outcomes required were change in depressive symptoms and number of dropouts from study or treatment (as a measure of acceptability). Data collection and analysis We extracted data, assessed risk of bias in duplicate, and resolved disagreements through discussion or consultation with a third person. We conducted random-effects meta-analyses when appropriate. We summarised continuous outcomes using mean differences (MDs) or standardised mean differences (SMDs), and dichotomous outcomes using risk ratios (RRs). Main results We included six trials (n = 698; most participants were women approximately 40 years of age). All studies evaluated psychotherapy plus usual care (with antidepressants) versus usual care (with antidepressants). Three studies addressed the addition of cognitivebehavioural therapy (CBT) to usual care (n = 522), and one each evaluated intensive short-term dynamic psychotherapy (ISTDP) (n = 60), interpersonal therapy (IPT) (n = 34), or group dialectical behavioural therapy (DBT) (n = 19) as the intervention. Most studies were small (except one trial of CBT was large), and all studies were at high risk of detection bias for the main outcome of self-reported depressive symptoms. A random-effects meta-analysis of five trials (n = 575) showed that psychotherapy given in addition to usual care (vs usual care alone) produced improvement in self-reported depressive symptoms (MD-4.07 points, 95% confidence interval (CI)-7.07 to-1.07 on the Beck Depression Inventory (BDI) scale) over the short term (up to six months). Effects were similar when data from all six studies were combined for self-reported depressive symptoms (SMD-0.40, 95% CI-0.65 to-0.14; n = 635). The quality of this evidence was moderate. Similar moderate-quality evidence of benefit was seen on the Patient Health Questionnaire-9 Scale (PHQ-9) from two studies (MD-4.66, 95% CI 8.72 to-0.59; n = 482) and on the Hamilton Depression Rating Scale (HAMD) from four studies (MD-3.28, 95% CI-5.71 to-0.85; n = 193). High-quality evidence shows no differential dropout (a measure of acceptability) between intervention and comparator groups over the short term (RR 0.85, 95% CI 0.58 to 1.24; six studies; n = 698). Moderate-quality evidence for remission from six studies (RR 1.92, 95% CI 1.46 to 2.52; n = 635) and low-quality evidence for response from four studies (RR 1.80, 95% CI 1.2 to 2.7; n = 556) i...