In an extensive review of more than 400 hypnotic treatment outcome studies, Flammer and Bongartz (2003) presented meta-analytic evidence supporting the efficacy of hypnosis as a therapeutic tool. Meta-analyses, however, are prone to the problem of selective publication of studies reporting positive outcomes. In the present investigation, we therefore employed a variety of methods to test for the presence of publication bias in the data analysed by Flammer and Bongartz (2003). The results suggest that publication bias may have contributed to the effect size estimate by about one third. However, our analysis also shows that the efficacy of hypnosis is of a substantive nature, and may not be explained on the basis of publication bias alone. Copyright © 2008 British Society of Experimental & Clinical Hypnosis. Published by John Wiley & Sons, Ltd.