Color serves diverse functions in mammalian hairs, including crypsis, sexual display, and resistance to wear. Our understanding of the roles that these colors have played in mammalian evolution has been hindered by both a limited understanding of their mechanistic bases and a complete inability to know the colors of mammals known only from fossils. We therefore propose to collect data on melanin (and potentially other pigments) in extant mammals and then use them to reconstruct colors of fossil mammals from China and thereby test hypotheses on melanin-based color evolution. First, we propose to enhance our understanding of the form-function relationships in melanin-based colors in extant mammals through collection of extensive data on melanosome morphology, chemistry, and hair color. We will then use these data to address hypotheses about the relationships between morphology and color while laying the groundwork for analysis of fossil data. Second, we bring the methods from the first section together to reconstruct colors and color patterns of fossil mammals and address hypotheses about the diversification of melanosome morphology. This novel research proposal thus flows from proximate to ultimate, and has the potential to transform our understanding of a critical part of the structure and physiology of organisms, the integument. Our extensive preliminary data and proven record of collaborative success show that the project is feasible and has a high probability of success.