The EU member states are required to designate Natura2000 protected sites and to ensure that these form an ecologically coherent network. However, no assessment of coherence in the terrestrial environment has yet assessed all the core attributes important for ecological coherence (representivity, replication, patch sizes and shapes, and connectivity). In this study, a method is presented that addresses all of the above criteria for three main Annex I habitats in Scotland (Caledonian woodland, Tilio-acerion forest of slopes, screes and ravines, and Northern Atlantic Wet heath with Erica tetralix). The spatial distributions of these habitats were modelled based on knowledge rules identified from the British National Vegetation Community (NVC) descriptions. Only knowledge rules that were shown to significantly improve the detection of individual habitats were applied. Representivity and replication were subsequently assessed by determining the area and number of habitats within the Natura2000 sites. Size and shape of the habitat patches were analysed with statistical measurements. The connectivity was modelled with least-cost networks. Two sources of costs were used and they were assigned to four different land-cover maps. Results of these analyses were reported for the whole of Scotland and for Scotland Rural Development Programme’s Regional Project Assessment Committees regions. 202,648 ha of Caledonian pinewood, 252,216 ha with ravine woodland and 1,551,805 ha of wet heath were detected. Validation, based on NVC surveys, showed the detection to range between 34% and 61% with correct classification. More than double the amount of area were found outside Natura2000 protected areas. The patches were generally small and their shapes not compact. The connectivity between the sites varied significantly depending on the costs and land-cover to which costs were assigned. These results provide a good base-line for the assessment of ecological coherence, but more analysis is needed to fully know whether the sites form a coherent network.