For memory to be efficient and useful during recall, problem-solving, and planning, retrieval must be compressed in time. Evidence from rodents suggests that neural compression during replay of spatial memories varies widely, with a range of compression ratios reported from 6:1 to 64:1. Anecdotal evidence suggests that similar compression occurs during mental navigation in humans: we recall how to get from one place to another countless times almost every day of our lives, and this recall never takes as long as physically travelling those routes would take. In this experiment we sought to determine whether this behavioural compression could be measured during mental navigation in humans (spatial memory replay), and which factors might affect the compression of such spatial memories. To this end, thirty participants mentally navigated routes between two landmarks, which varied in length and number of turns, as we measured replay times and recorded ratings of familiarity, detail, and presence. A multi-level model was used to determine which factors were associated with variation in compression. Route length and number of turns emerged from this model as significantly correlated with compression, such that longer routes were more compressed while compression was attenuated as the number of turns in a route increased. This suggests that compression during recall may be affected by specific features of a route, especially those that may act to segment the space or event being represented.