Visually impaired people (VIP) have to rely on different information to generate a cognitive map of their environment than normally sighted people. This study explored the extent to which a cognitive map could be generated by auditory information of route-type and survey-type descriptions of a fictitious environment. A total of 27 visually impaired and 28 normally sighted participants listened to either a survey-type or a route-type description of a fictitious zoo. They then answered both route-type and survey-type questions. This listening/question-and-answer sequence was repeated twice (total n ¼ 3). The visually impaired participants showed no difference in error frequency between the two description types, while the normally sighted individuals performed better after listening to the survey-type description. In addition, the learning curve of the normally sighted individuals was steeper than that of the visually impaired and they made fewer errors. The error scores indicated two subgroups in both the normally sighted and the visually impaired groups. These two groups, the 'good' learners and the 'poor' learners, showed marked differences in generating a cognitive map from auditory descriptions of an environment.