Background: Successful navigation is crucial to everyday life. Individuals with Williams syndrome (WS) have impaired spatial abilities. This includes a deficit in spatial navigation abilities such as learning the route from A to B. To-date, to determine whether participants attend to landmarks when learning a route, landmark recall tasks have been employed "after" the route learning experience. Here, we combined virtual reality and eye tracking technologies, for the first time, to measure landmark use in typically developing (TD) children and participants with WS "during" route-learning. Method: Nineteen individuals with WS were asked to learn a route in a sparse environment (few landmarks) and in a rich environment (many landmarks) whilst their eye movements were recorded. Looking times towards landmarks were compared to TD children aged 6, 8 and 10 years. Changes in attention to landmarks during the learning process were also recorded. Results: The WS group made fewer looks to landmarks overall, but all participants looked for longer at landmarks that were at junctions and along the paths of the maze than landmarks that were in the distance. Few differences were observed in route learning between the sparse and rich environments. In contrast to the TD groups, those in the WS group were as likely to look at non-unique landmarks as landmarks at junctions and on paths. Discussion: The current results demonstrate that attention to landmarks during route learning reflects the types of landmarks remembered in memory tasks, that individuals with WS can learn a route if given sufficient exposure, but that this is accomplished within the context of an impaired ability to select appropriate landmarks.