Consumption taxes are often considered as the most anti-redistributive component of the tax system. Yet, very few estimates, and fewer international comparisons of the redistributive impact of consumption taxes exist in the literature, due to scarce data on household expenditures. We use household budget surveys and microsimulation to provide consistent estimates of the regressivity of consumption taxes for a large panel of countries and years. We propose a new method for imputing consumption expenditure across countries, using widely available data on income and socio-demographic characteristics of households. We show that including the distribution of housing rents, when data is available, to impute households' consumption greatly improves the prediction of the model. Our results are threefold. First, there is a 1 to 2 ratio between the propensity to consume of the top decile (around 50% of their income) and that of the bottom decile (100% of income). Second, consumption taxes entail a significant rise in the Gini coefficient of income (between 0.01 and 0.04 point), yet of much smaller magnitude than the positive redistribution operated by direct taxes and transfers. Third, cross-country differences in the distributive effect of consumption taxes are mainly explained by variations in the tax rate (from 7 to 24% in our sample), rather than variations in the distribution of consumption, since everywhere the propensity to consume declines sharply with income.