Background: Effect measures for meta-analyses of continuous outcomes include mean differences (MD), standardized MD (MD in pooled standard deviation units, SMD), and ratio of (arithmetic) means (RoM). Recently, ratio of geometric means using either ad hoc (RoGMadhoc) or Taylor series (RoGMTaylor) methods for estimating variances have been proposed to pool skewed data. Objectives: To compare treatment effects and heterogeneity of RoGMadhoc and RoGMTaylor to MD, SMD, and RoM using empiric data and simulation. Methods: We searched the Cochrane Database of Systematic Reviews (2008, Issue 1) for reviews reporting continuous outcomes; from each review we selected the meta-analysis with the most (and at least 5) trials. Meta-analyses were conducted using each effect measure. Pairwise differences between methods in treatment effect and heterogeneity (Cochrane s Q) p-values, compared separately, were assessed using non-parametric sign tests, and assymetry of discordant pairs (statistically significant result for only one of two effect measures) using Exact tests. Simulation parameters employing both normal and log-normal distributions were chosen to be representative of those commonly encountered in meta-analyses of continuous outcomes. Results: 232/5053 systematic reviews met inclusion criteria. RoGMTaylor exhibited similar treatment effects, with more heterogeneity (p ≤ 0.0001-0.008) for distributions with median coefficient of variation (CV)≤0.6, and less heterogeneity (p=0.0002-0.0004) for distributions with median CV≥0.6 when compared to MD, SMD and RoM. RoGMadhoc exhibited more extreme treatment effects (p≤0.0001 0.17) and greater heterogeneity (p≤0.0001) compared to all other effect measures. Most heterogeneity comparisons also demonstrated statistically significant asymmetry of discordant pairs. Simulation confirmed these results. Conclusions: RoGM using Taylor series to estimate the variance may be considered for pooling continuous outcomes of skewed data in which CV is high. However, given clinicians lack of familiarity with geometric means and acceptable performance characteristics of RoM in most situations, RoM is a reasonable alternative ratio method for pooling continuous outcomes even in such cases.