This paper addresses the question of whether the subject prefix in Nairobi Swahili is an agreement marker, a pronoun, or ambiguous between the two, as has been suggested previously for other varieties of Swahili and other Bantu languages. Bresnan and Mchombo (1987) show that in Chichewa, the subject prefix functions both as a pronoun as well as an agreement marker, while Keach (1995) shows that in Standard Swahili the subject prefix is also ambiguous between a pronoun and an agreement marker. This paper develops Keach's analysis of the subject prefix in a regional dialect of Swahili spoken in and around Nairobi, the capital of Kenya. Various syntactic tests of subject-hood and topicality are applied to the subject prefix: whether constructions involving the subject prefix occur in quantified contexts, as answers to wh-questions, and as topicalized idiom chunks. The results of these tests, as well as additional typological evidence, show that the subject prefix in Nairobi Swahili is a pure agreement marker, and bears no pronominal or subject-like functions. The discrepancies between the current study and that of Keach are attributed to dialectal and grammaticalization differences.