In recent years, theorists of social inequality have increasingly rejected analytic models using the individual as the unit of analysis, favoring "relational" models centered on the dynamic, group-level interactions that can account for disparities in the distribution of job rewards. In this article we scrutinize three distinct strands of relational thinking: categorical theories, analysis of symbolic boundaries, and theories of intersectionality. Our goals are twofold: first, to identify some of the major conceptual and methodological limitations in this field and, second, to begin the task of conjoining insights from each strand of thinking, fostering conceptually richer and more powerful theoretical formulations of organizational inequalities. The article sketches some potential avenues for empirical analysis that seem likely to advance relational models and highlight their advantages—advantages that provide rich sociological guideposts—compared to more individually centered or even aggregate descriptive models that have governed the field since World War II.