This paper reexamines the determinants of redistribution in light of improved data and methods relative to earlier literature. In particular, we use the latest version of the UNU-WIDER's Income Inequality Database to have the best available estimates of both pre- and post-redistribution inequality for the largest set of countries and periods. We tackle head-on problems related to model specification that risk generating large biases in estimates because of mechanical associations between the dependent and explanatory variables. The results suggest that the bias in the earlier work can be substantial. The descriptive analysis highlights, in addition, how scarce the data are when it comes to information about the extent of redistribution in developing countries.