This meta-analysis looked at 17 studies which focused on the comparison of reading on screen and reading on paper in terms of reading comprehension and reading speed. The robust variance estimation (RVE)- based meta-analysis models were employed, followed by four different RVE meta-regression models to examine the potential effects of some of the covariates (moderators) on the mean differences in comprehension and reading speed between reading on screen and reading on paper. The RVE meta-analysis showed that reading on paper was better than reading on screen in terms of reading comprehension, and there were no significant differences between reading on paper and reading on screen in terms of reading speed. None of the moderators were significant at the 0.05 level. In the meanwhile, albeit not significant, examination of the p-values for the difference tests prior to 2013 and after 2013 respectively (not shown here) indicated that the magnitude of the difference in reading comprehension between paper and screen followed a diminishing trajectory. It was suggested that future meta-analyses include latest studies, and other potential moderators such as fonts, spacing, age and gender. •Reading on paper is better than reading on screen in terms of reading comprehension.•Reading on paper is not significantly different from reading on screen in terms of reading speed.•The magnitude of the difference in reading comprehension between media types followed a diminishing trajectory.