The study investigates the relationships between working memory (WM), amount of learning, and strategies used in spatial description. WM involvement and strategies reported were assumed to change, depending on whether the text learning was extensive or limited. Two experiments were carried out using dual-task paradigm: participants listened to spatial text three times and concurrently one group performed a spatial concurrent task, one group a verbal task (to measure WM involvement), and one group no secondary task. In Experiment 1, participants listened three consecutive times then performed recall tasks (one verbal – verification test; one spatial – graphical representation). In Experiment 2, recall tasks were performed after first and third listening. The strategies used were ascertained through a questionnaire. Results showed that the verification test was impaired by the verbal concurrent task after listening three times (Experiment 1) and after first listening (Experiment 2). The graphical representation performance was impaired by verbal and spatial concurrent tasks, detected only after listening three times, not after a single time (Experiments 1 and 2). The strategies most used were visuo-spatial; their relationship with WM changes as a function of number of times of listening. Overall, the results showed that extensive learning allows construction of a spatial mental representation that is modulated by WM and strategies.