Abstract In this paper we apply meta-analysis to a sample of 27 empirical studies to clarify the association of board independence and ownership concentration with voluntary disclosure. We examine whether variations in results are attributable to the differences in the corporate governance system, the investor protection rights and the measurement of the governance variables. The findings show that the positive association between board independence and voluntary disclosure only occurs in those countries with high investor protection rights. The findings emphasize the need to consider the legal and institutional setting explicitly when analysing the effect of corporate governance on voluntary disclosure. Acknowledgements We would like to thank two anonymous referees, the guest editors of EAR for this special issue, Katherine Schipper and Marco Trombetta, and the seminar participants at the EAR Research Conference on 'Measurement Issues in Financial Reporting' (Segovia) for their helpful comments and suggestions. We also thank the Research Agency of the Spanish Government for financial support (Project SEJ2007-61450/ECON). Notes This is one reason why the number of studies and sample size of the overall meta-analysis does not agree with the sum of the effect sizes and samples of the subgroup analyses. The other is that there is also one study (Cerbioni and Parbonetti, Citation2007) whose sample consists of 10 different countries, so it is included in the overall meta-analysis but not in the subgroup analyses. Nevertheless, our findings do not change if we discard this study. In the concentration analysis we also lose one study because the data in La Porta et al. Citation(1998) is not provided for Kenya. It could be argued that these file drawer papers are unlikely to contain reliable results because they would not be in the file drawer otherwise, but this is just a theoretical assumption. In contrast, Pomeroy and Thornton Citation(2008) assert that publication bias is particularly acute in accounting due to the absence of replication studies on the same topic. The separation of outside investor rights between high and low is made considering as low those countries with a score from 0 to 4, and high those with 5. We also check the robustness of results to considering low outside investor rights those countries with scores from 0 to 3 (4 studies) and high those with 4 to 5 (14 studies) and the results do not change. For legal enforcement, we consider high those above the median and low those below the median. Since there are studies with the same score in legal enforcement and equal to the median, we also check the robustness of results to inclusion in one or the other group, and the results do not change. We consider the same criteria as in board independence to divide between high and low outside investor rights and legal enforcement. We also check the robustness of results to considering as low outside investor rights scores between 0 and 3 (7 studies) and high those with scores between 4 and 5 (11 studies) and the results remain unchanged. The results are also robust to inclusion in legal enforcement of those studies with scores equal to the median in one group or the other.