To date, meta-analyses of effects of acidification have focused on the overall strength of evidence for statistically significant responses; however, to anticipate likely consequences of ocean acidification, quantitative estimates of the magnitude of likely responses are also needed. Herein, we use random effects meta-analysis to produce a systematically integrated measure of the distribution of magnitudes of the response of coral calcification to decreasing ΩArag. We also tested whether methodological and biological factors that have been hypothesized to drive variation in response magnitude explain a significant proportion of the among-study variation. We found that the overall mean response of coral calcification is ~15% per unit decrease in ΩArag over the range 2 < ΩArag < 4. Among-study variation is large (standard deviation of 8% per unit decrease in ΩArag). Neither differences in carbonate chemistry manipulation method, study duration, irradiance level, nor study species growth rate explained a significant proportion of the among-study variation. However, studies employing buoyant weighting found significantly smaller decreases in calcification per unit ΩArag (~10%), compared with studies using the alkalinity anomaly technique (~25%). These differences may be due to the greater tendency for the former to integrate over light and dark calcification. If the existing body of experimental work is indeed representative of likely responses of corals in nature, our results imply that, under business as usual conditions, declines in coral calcification by end-of-century will be ~22%, on average, or ~15% if only studies integrating light and dark calcification are considered. These values are near the low end of published projections, but support the emerging view that variability due to local environmental conditions and species composition is likely to be substantial.