Previous studies of preference for and harmony of color combinations have produced confusing results. For example, some claim that harmony increases with hue similarity, whereas others claim that it decreases. We argue that such confusions are resolved by distinguishing among three types of judgments about color pairs: (1) preference for the pair as a whole, (2) harmony of the pair as a whole, and (3) preference for its figural color when viewed against its colored background. Empirical support for this distinction shows that pair preference and harmony both increase as hue similarity increases, but preference relies more strongly on component color preference and lightness contrast. Although pairs with highly contrastive hues are generally judged to be neither preferable nor harmonious, figural color preference ratings increase as hue contrast with the background increases. The present results thus refine and clarify some of the best-known and most contentious claims of color theorists.