We propose that people can and will infer group memberships from resource distributions, and that these distributions have implications for people's understandings of the groups themselves and their own associations with these groups. We derive hypotheses from social identity and self-categorization theories, and test them in three experiments. In Experiment 1, participants systematically rated specific patterns of group memberships as more likely than others in light of specific resource distributions in a manner consistent with our predictions. In Experiment 2, intragroup distributive fairness led to greater perceived self-in-group similarity than intra-group distributive unfairness, while distributively unfair, in-group favouritism led to greater perceived self-in-group similarity than intergroup fairness. In Experiment 3, social identification dropped following unfair, out-group favouritism and intragroup unfairness, but not unfair, in-group favouritism, or intragroup and intergroup fairness. The current data provide support for our hypotheses and clear evidence that resource distributions can be providers of group membership information.