This research applies GIS to the analysis of a previously collected dataset of 131 macroinvertebrate samples from headwater streams (defined as within 2.5km of source) on four rivers in England and Wales. The particular emphasis of the analysis is the influence of land use factors on both species composition and biotic indices representing habitat quality and tolerance to pollution. In the GIS, a database of digital environmental datasets was assembled, including rivers, soils, solid and drift geology, elevation, and land use/cover information from both field surveyed and remotely sensed sources. The GIS was used to introduce new variables into the analysis, particularly at the catchment scale, in addition to field measured habitat characteristics. Surface and hydrological modelling functions in ARC/INFO and in purpose-written C programs were used to derive catchment boundaries and to generate Euclidean and upslope buffer zones, within which variables were extracted and quantified. This enabled the influence of land use in close proximity to the streams to be compared to catchment scale land use. Standard multivariate statistical techniques were applied to look for relationships among the environmental and biological variables. Correspondence analysis (CA) and canonical CA methods revealed that catchment and site scale variables could explain a small proportion of the variance in species composition but that the land use of variables were not directly important. Some weak but significant associations between the biotic indices and particular land use factors, such as proportions of bracken and suburban land cover, and the average contagion of landscape patches, were found by multiple regression analysis. The results suggested that catchment scale land use was more important than within the buffers, although the differences were marginal.