Simulated Thematic Mapper (TM) data are examined for an area of the North Yorks Moors, northern England. The aim of analysis is to determine the number and identity of wavebands needed to distinguish specific moorland vegetation types from the surrounding community. The interband correlation and dimensionality of the data are found to be different for each of the vegetation types studied. Calculation of the transformed divergence measure shows that the major communities of bracken, heather and sedges are clearly separated with the use of one or two wavebands; four or five bands are needed to distinguish the stages of heather growth. In all cases, the combination of wavebands which maximizes discrimination is specific to the vegetation types to be separated. The work has practical implications for the use of TM data in the monitoring of the moorland landscape.