Fluctuating asymmetry, the random departure from perfect bilateral symmetry, is a common measure of developmental instability that has been hypothesized to be inversely correlated with heterozygosity. Although this claim has been widely repeated, several studies have reported no such association. Therefore, we test the generality of this association, using meta-analysis, by converting test statistics for the relationship between heterozygosity (H) and fluctuating asymmetry (FA) into a common effect size, the Pearson’s product-moment correlation coefficient. We have analysed a database containing 41 studies with a total of 118 individual samples. Overall we found an unweighted mean negative effect size; r=−0.09 (i.e. a negative correlation between H and FA). Significant heterogeneity in effect size was mainly caused by a difference between ectothermic and endothermic animals, and to a lesser extent by the use of different study designs (i.e. within-population vs. among-populations). Mean effect size for endothermic animals was positive and significantly different from the mean effect size for ectothermic animals. Only for within-population studies of ectothermic animals did we find a significantly negative effect size (r=−0.23 ± 0.09). The distribution of effect sizes in relation to sample size provided little evidence for patterns typical of those produced by publication bias. Our analysis suggests, at best, only a weak association between H and FA, and heterozygosity seems to explain only a very small amount of the variation in developmental instability among individuals and populations (r2=0.01 for the total material).