Colour, derived primarily from melanin and/or carotenoid pigments, is integral to many aspects of behaviour in living vertebrates, including social signalling, sexual display and crypsis. Thus, identifying biochromes in extinct animals can shed light on the acquisition and evolution of these biological traits. Both eumelanin and melanin-containing cellular organelles (melanosomes) are preserved in fossils, but recognizing traces of ancient melanin-based coloration is fraught with interpretative ambiguity, especially when observations are based on morphological evidence alone. Assigning microbodies (or, more often reported, their 'mouldic impressions') as melanosome traces without adequately excluding a bacterial origin is also problematic because microbes are pervasive and intimately involved in organismal degradation. Additionally, some forms synthesize melanin. In this review, we survey both vertebrate and microbial melanization, and explore the conflicts influencing assessment of microbodies preserved in association with ancient animal soft tissues. We discuss the types of data used to interpret fossil melanosomes and evaluate whether these are sufficient for definitive diagnosis. Finally, we outline an integrated morphological and geochemical approach for detecting endogenous pigment remains and associated microstructures in multimillion-year-old fossils.