Background Attention on harmful effects of psychological interventions for adult post-traumatic stress disorder (PTSD) has increased, yet a comprehensive meta-analysis is lacking. Aims To summarise incidences and relative risks of deterioration, adverse events (AEs) and serious adverse events (SAEs) in trials of psychological interventions for adult PTSD. Method We searched MEDLINE, PsycInfo, Web of Science and PTSDpubs from inception to 21 April 2022 for sufficiently large (n ≥ 20) randomised controlled trials (RCTs) reporting on the incidence of harms. Results We included 56 RCTs (4230 patients). Incidences of harms were generally low (0–5%). Psychological interventions were associated with decreased risk of deterioration relative to passive (RR = 0.21, 95% CI 0.15–0.28) and active control conditions (RR = 0.36, 95% CI 0.14–0.92). Decreased risk was even more pronounced in sensitivity analyses on trials exclusively delivering treatments face to face. When compared with other psychological interventions, trauma-focused cognitive–behavioural therapy (TF-CBT) was associated with decreased risk of SAEs (RR = 0.54, 95% CI 0.31–0.95) and with no differential risk of deterioration and AEs. Conclusions The current evidence base suggests that psychological interventions are safe for most adults with PTSD. In none of the analyses were psychological interventions associated with an increased risk of harm compared with control conditions. TF-CBT was found at least as safe as other psychological interventions. Individual face-to-face delivery might be the safest delivery format. However, more data are needed to draw firmer conclusions. We encourage research teams to routinely and thoroughly assess and report the incidence of harms and their causes.