A model for sentence memory is described that extracts a parsed surface representation of the sentence and a propositional representation of a sentence. In addition, if possible, pointers are added to a past situational referent for the sentence. This system predicts basic results in sentence memory without assuming different retention functions for surface, propositional, or situational information. There is better retention for gist than surface information (Anderson, 1974; Zimny, 1987) because of the greater complexity of the surface representations. Situational information is still better retained (Reder, 1982; Zimny, 1987) because of the greater practice of the situational referent. This model's only inference during sentence comprehension is to insert a pointer to an existing situational referent. Nonetheless, it is capable by this means of modeling many effects attributed to inferential processing (Reder, 1982; Schustack & Anderson, 1979; Bower, Black, & Turner, 1979). Equally important, this model is implemented in the ACT-R architecture (Anderson & Lebiere, 1998) that has been used to account for a wide variety of results in list memory (Anderson, Bothell, Lebiere, & Matessa, 1998). In addition to establishing that sentence memory is like other memory, the ACT-R architecture provides a mechanism for mixing the various memory strategies that subjects bring to bear in these experiments.