Consider two similarity facts: a is similar to b with respect to G, and c is similar to d with respect to G. According to the Platonist approach to similarity, the analysis of such facts forces us to admit that similarity facts are to be analyzed into facts about universal similarities of the form: a is similar to b with respect to G, and c is similar to d with respect to G, where similarity is a universal. In this paper, (...) I defend Neutralism, a view according to which there are properties and similarities, but these are not universals, particulars, or of other categories. After presenting the Platonist analysis of similarity proposed by Russell and Grossmann, I examine the question of whether or not the Platonist analysis of similarity leads us inevitably to a conception of similarity as a universal. Then I offer a novel, alternative analysis of similarity free of commitment to universals. (shrink)