Coastal Marind has a gender system comprising four genders. I follow the conventions established by Drabbe (1955) and refer to them as Genders I-IV (only replacing Drabbe’s Arabic numerals with Roman ones). The vast majority of nouns that I have recorded are conventionally assigned to one of the four genders, and always trigger agreement according to that gender. The basis for gender assignment is animacy: all nouns denoting animate beings are assigned to Genders I and II, and inanimates to Genders III and IV. Within animates, Gender I contains male humans, while Gender II contains female humans and all animals (regardless of biological sex). In contrast, there is no obvious rationale (semantic or phonological) for the assignment of inanimates to Gender III and IV, but note that Gender III is considerably larger (452 recorded members) than Gender IV (195 members). There are also good reasons to consider Gender III a ‘default’ gender, because when agreeing targets (e.g. demonstratives) occur in syntactic contexts that do not allow agreement (e.g. in some adverbial positions) it is always the Gender III form of the target that is used. Similar four-gender systems - featuring roughly one masculine, one feminine and two inanimate genders - are found throughout the Anim language family. Usher and Suter (2015) present data (from missionary-linguists Roland Fumey and Robert Petterson) showing evidence of four genders in the languages Kuni (of the Lake Murray subgroup) and Ipiko (of the Inland Gulf subgroup), and ongoing research by Phillip Rogers has unearthed four genders in Bitur (of the Lower Fly subgroup). During my own fieldwork I was also able to identify four genders in the other languages of the Marindic subgroup, i.e. Bush Marind and the Upper Bian language. It seems reasonable to conclude that four genders were present in the proto-Anim language and that they have been inherited - with some differences in assignment principles - by its present-day descendants. Like Corbett (1991), I find Hockett’s definition of gender as “classes of nouns reflected in the behaviour of associated words” (1958: 231) to be a useful starting point for the investigation. There is a small number of nouns in Coastal Marind that exhibit vowel alternations corresponding to assignment to different genders (with a predictable change in meaning), but it is the pervasive phenomenon of gender agreement in targets such as determiners, adjectives and verbs that makes it necessary to posit the category gender for Coastal Marind. I provide an overview of the reflection of gender in such ‘associated words’ in §5.1. The principles of gender membership are reviewed in more detail in §5.2. More detailed discussion of some agreement phenomena is in §5.3.