This Masters thesis aims to determine the requirements of a semantics for imperatives in the domain of navigation instructions. Pragmatic and semantic theories and computational approaches to imperatives will be examined. The overall aim is to bridge the gap between natural language imperatives and their interpretation in a logic of action, and to determine what is required to represent a range of linguistic phenomena from a corpus of navigation instructions. A grammar providing an underspecified semantic representation of simple imperatives is presented and issues of what type of interpretation should  be performed at which level in a hypothetical architecture for interpreting imperatives are discussed.