The focus of this paper is the syntactic deficit in agrammatic aphasia. The specific issue is the extent to which prepositions are impaired in this syndrome. This category is of particular interest because of the unique role its members play in the grammar. This is the organization of the paper: First, several descriptive generalizations are examined critically, and arguments against them are advanced. Then, a new hypothesis is formulated, stated in terms of current linguistic theory. This hypothesis views the deficit as being partial from a syntactic point of view. The relevant notion to account for the data (i.e., partitioning the impaired from preserved prepositions) is Government, a structural relation that must hold between the preposition at issue and the verb. The consequences of this hypothesis are derived, and an experiment that was conducted to test them is reported. The findings of this experiment not only support the hypothesis, but also suggest that the impairment is unique to agrammatic aphasic patients, since the performance of a control group of fluent aphasics was different. Finally, several theoretical issues are discussed in light of the findings and of the proposed description of the agrammatic deficit.