Preferences over income distribution are the basis for a variety of models that aim at explaining results in economic experiments. The direct evidence concerning these preferences, however, is limited to a relatively small set of games. The authors discuss crucial evidence, including that from a recent large-scale Internet experiment and the different models' performance. It appears that subjects are highly heterogeneous and that the relative impact of different motives depends on a variety of factors. The authors present a general model that can help as a starting point for measuring distributional preferences.