The phenomenon of prodrop has attracted much attention in recent years both in the field of syntactic theory and in pragmatics (Chomsky, 1981, 1982; Rizzi, 1982; Gundel, 1980; Givon, 1976). Persian is among the many languages which exhibit this phenomenon. The central aim of this paper is to discover the pragmatic contraints governing the distribution of overt and non-overt subjects. Persian is a SOV language. Verbs carry the person and number information in the form of suffixes. These agreement markers are unique for each person. Non-overt (zero) subjects are allowed and occur very frequently in speech. Looking at natural data, we see that overt subjects appear even when their referents are in focus. Finding the contraints governing the distribution of these overt subjects is the primary question addressed in this paper. For the purposes of this study, we are classifying full noun phrases and pronouns together as overt subjects although pronominalization has its own separate pragmatic functions. Here we are focussing on zero subjects and overt subjects. Our data base consists of six children and four adult narratives. The unity of the narrative as a genre with its usual focus on one main actor make it possible to track the same noun phrase throughout the story in a variety of syntactic and pragmatic environments. This makes the narrative genre ideal for contrasting overt and non-overt subjects.