Space syntax research has had significant success over the years and has served to illustrate the importance of configurational measures, especially those that take into account all the spaces in a system. Here, assumptions of axial lines as elementary spatial units have been overwhelming. Although based on a theoretical construct of visibility, this postulation has rarely been extended to perception or cognition and this has given rise to questions about geometric and metric considerations. The research presented here was carried out in three large urban hospitals. In them, 128 volunteers performed 'open searches, where they attempted to become familiar with the hospital; 'directed searches', where they sought specific locations; and various cognitive mapping tasks such as pointing to locations that were not within sight and sketching the main corridors and routes of the hospital. The hospitals themselves were analyzed through conventional syntax measures of axial lines and a segmented version of those lines. Correlational and regression analyses revealed that, although use of space was best predicted by local measures (connectivity), distribution of people was better explained through integration-3. Performance comparisons between the environmental measures given by 'whole' and 'segmented' lines suggested that, at least in complex architectural settings, original syntax definitions of axial lines as uninterrupted visibility lines have more predictive power and better cognitive presence than 'segmented' lines. The results also support previous findings of intelligibility as a factor in predicting space use by extending it to the cognitive realm. Additionally, it has brought out the reverse role of intelligibility-3. It was also found that mean depth of an entry effects the way a building is explored; therefore, knowledge of this spatial property may provide a fair indication of the relative importance which each space will have on wayfinding.