This paper aims to conduct a comparison of the efficiency of Islamic banks (IBs) versus conventional banks (CBs) based on meta-analysis. Empirical studies in this field, mainly in MENA or Gulf Cooperation Country regions, usually provide mixed conclusions and we do not know the reasons of this discrepancy in the results.We conduct a meta-regression analysis (MRA) that is based on 35 studies and includes 484 estimates of efficiency scores by bank type among other characteristics. We employ Bayesian averaging technique to identify the main covariates explaining the heterogeneity of the results. Our findings suggest, first, no evidence of the superiority of one banking system over the other, neither in terms of technical, cost or profit efficiency. Second, we identify several important bank's and sample characteristics that have also important impacts on the provided efficiency estimates. Third, when we compare the studies into those who conduct a deep statistical comparison of the efficiency distributions by bank type and those who omit to do it, we find that CBs outperform IBs, suggesting a potential publication bias. Finally, with respect to the inefficiency, MRA suggests that the regions banking systems suffer from the inability of banks to maximize profit, so they are much less profitable than it should be. There are also some managerial deficiencies linked to cost minimization, rendering the region's