Word order in Itelmen (itl, Chukotko-Kamchatkan) is quite flexible and has not been previously studied in any depth. This talk draws on a small text corpus to investigate the role of Information Structure in conditioning the distribution of object-verb and verb-object (OV,VO) word orders. Although speakers in elicitation contexts generally assent to both orders and describe them as meaning the same, a robust pattern emerges in the corpus: O denoting new discourse entities are overwhelmingly preverbal, while given objects may occur pre- or post-verbally. In this talk, I argue that these results have a variety of implications: (i) the observed pattern converges with other evidence that Itelmen is an OV language and provides an argument that the VO order in Itelmen is not simply a calque from Russian, (ii) the Information-Structural evidence provides a means to resolve a syntactic puzzle about the analysis of perception verb complements in Itelmen, (iii) the pattern contributes to larger debates about the syntactic/grammatical representation of new versus given (or topic and focus), potentially arguing against “focus-movement” (cartographic) perspectives, and (iv) Information Structure provides a better characterization of the OV/VO alternations than competing accounts of such alternations that appeal to extra-grammatical communicative efficiency, notably those rooted in ambiguity avoidance with animate objects.