A meta-analysis of treatments for nightmares is reported. The studies were identified by database searches and by an inspection of relevant reference lists. The inclusion criteria were: nightmares as a target problem, studies published in English, use of a randomized controlled trials and reporting of nightmare-relevant outcomes. A total of 19 studies, published between 1978 and 2012 were identified, which included 1285 participants. Effect sizes were calculated as Cohen’s d. A statistically significant improvement for all studies combined (d = 0.47, 95% CI = 0.33–0.60, fixed effects model; d = 0.49, 95% CI = 0.32–0.66, random effects model) and for psychological treatments alone (d = 0.48, 95% CI = 0.36–0.60, random) and for prazosin alone (d = 0.50, 95% CI = 0.03–0.96, random) was found. Individual therapy format yielded a higher effect size than a self-help format (p = 0.03). Minimal interventions (relaxation, recording) yielded lower overall effect size than studies offering more extensive interventions (p = 0.02). It is concluded that there are both psychological and pharmacological interventions which have documented effects for the treatment of nightmares.