Three experiments investigated whether the spatial distance effect in situation models (Rinck & Bower, 1995) is based on direct distance or on route distance. In all experiments, participants learned the layout of a building before they read narratives about a protagonist's motions and actions in this building. Reading times for target sentences referring to objects located in different rooms of the building were measured. Varying (a) the direct distance between the target object and the protagonist's current location (location room) and (b) the length of the route the protagonist walked from the target object to the location room yielded a direct distance effect (increased reading times with increased direct distance). Qualitatively, this effect was robust against manipulation of the layout learning procedure and the experimental narratives, although it decreased in magnitude. Assuming that variation of route length implies variation of the temporal distance between the protagonist leaving the target object and arriving in the location room, readers were then instructed to monitor the narrative time course. Under this instruction, a route distance effect was observed (increased reading times with increased route length). Results are discussed with regard to the organization of situation models.