OBJECTIVE To review the literature on the comparative efficacy of psychological, behavioural and medical therapies for acrophobia (fear of heights). METHODS Multiple databases were searched through the Cochrane Common Mental Disorders review group on 1 December 2015. RESULTS The data were extracted independently and were pooled using RevMan version 5.3.5. The main outcome measures were changes from baseline on questionnaires for measurement of fear of heights, such as the Acrophobia Questionnaire (AQ), Attitude Towards Height Questionnaire (ATHQ), and behavioural avoidance tests. Individual and pooled analyses were conducted. Sixteen studies were included. Analysis of pooled outcomes showed that desensitisation (DS) measured by the post-test AQ anxiety score (standardised mean difference [SMD], -1.24; 95% CI, -1.88 to -0.60) and in vivo exposure (IVE) were effective in the short term compared with control (SMD, -0.74; 95% CI, -1.22 to -0.25). IVE was not effective in the long term (SMD, -0.34; 95%CI -0.76 to 0.08) and there were no follow-up data for DS. Virtual reality exposure (VRE) therapy was effective when assessed with the ATHQ but not the AQ. Augmentation of VRE with medication was promising. The number needed to treat (NNT) ranged from 1.4 (95% CI, 1.0 to 2.2) for IVE therapy with oppositional actions (a psychological process) versus waitlist control to an NNT of 6.0 (95% CI, 2.8 to 35.5) for the rapid phobia cure (a neurolinguistic programming technique) versus a mindfulness exercise as the control activity. It was often unclear if there were biases in the included studies. CONCLUSIONS A range of therapies are effective for acrophobia in the short term but not in the long term. Many of the comparative studies showed equivalence between therapies, but this finding may be due to a type II statistical error. The quality of reporting was poor in most studies.