The book addresses grammatical gender in Romance, and its development from Latin. It works with the toolbox of current linguistic typology, and asks the fundamental question of how the Latin grammatical gender system gradually changed into those of the Romance languages. To answer this question, the book capitalizes on the pervasive dialect variation of which the better-known standard Romance languages only represent a fragment. Indeed, inspection of dialect variation across time and space forces one to dismiss the handbook account proclaiming that the neuter gender, contrasting with masculine and feminine in Latin, was eradicated from spoken Latin by late Empire times. Both Late Latin evidence and data from several modern dialects show that this never happened, and that the vulgate account proceeds from unwarranted back-projection of the data from modern languages like French and Italian. Rather, the neuter underwent transformations which are the main culprit for the differences in the gender system observed today between, say, Romanian, Sursilvan, Neapolitan, and Asturian, to cite just a few types of system which turn out to differ significantly. A precondition for establishing the database for diachronic investigation is a detailed description of many such systems, which reveals data whose interest transcends the diachronic issue under consideration: the book thus addresses systems where ‘husbands’ are feminine and others where ‘wives’ are masculine; discusses dialects where nouns overtly mark gender, but only in certain syntactic contexts; and proposes an analysis according to which one Romance language (Asturian) has split inherited grammatical gender into two concurrent systems.