When we examine a town map or view the town from a height we obtain survey knowledge, knowledge of the spatial configuration. When we navigate through the town we acquire procedural knowledge, knowledge about how our route connects different locations in the town. Three experiments investigated whether everyday cognitive maps formed after repeated navigation are primarily configurational or procedural when both configurational and procedural information on the routes travelled are available. If such maps are procedural then arguably they can be markedly inaccurate and lead to incorrect direction giving. The three experiments focussed on direction and location errors associated with constructing cognitive maps of routes from a town to a nearby university. Experiment 1 (N = 157), carried out when only one route was available, found that the distribution of the relative locations of the start and end points of the route from a frequentlynavigate-group's maps was markedly bimodal, including a large subset of similar but markedly inaccurate responses. In contrast, the distribution from an incidentally-map-experienced only group was unimodal and portrayed a high level of accuracy. Experiment 2 (N = 38) found that following the opening of a second route to the university navigators produced different patterns of directional errors across their cognitive maps of the old and new routes. Experiment 3 (N = 47) found that instructions to include a new detour in an otherwise straight street on the route led to statistically significantly more distortions of the street than instructions to ignore the new detour. The data are consistent with the view that everyday cognitive maps formed after repeated navigation are very often likely to be procedural, and also show a plasticity associated with the dominance in travelers' memories of what they do over what they view.