Variations in the composition of faunal assemblages in streams are often associated with changes in areas of substrate. An observer walking along the bank finds these areas easy to distinguish and we have used these as our basic ecological unit. Essentially the unit is visually distinct from the bank and we have called it a mesohabitat to introduce a scalar dimension which the term biotope does not have and to distinguish it from microhabitats such as a leaf tip or stone surface and macrohabitats which might include the whole of a reach of river. We wished to test the hypothesis that mesohabitats support specific faunal communities which remain distinct and faithful throughout the seasons and to explore the use of such information in environmental assessment. 10 replicate faunal samples, together with associated depth and velocity measurements, were collected from each of 8 mesohabitats ('Ranunculus fast', 'Ranunculus slow', 'Silt', 'Nasturtium', 'Phragmites', 'Sand', 'Gravel fast' and 'Gravel slow') in spring, summer and autumn from a lowland weeded stream, the Mill Stream, in southern England. Analysis of variance indicated that both abundance and faunal richness varied significantly between and within mesohabitats. The most valiability was noted in 'sand', 'Ranunculus slow', and 'silt'. Variation between mesohabitats was significant for all seasons but there was a tendency for the differences to be greatest in spring samples. Multivariate techniques, DECORANA and CCA (canonical correspondence analysis) showed that in general the mesohabitats selected for study remained distinct from one another but that the degree of separation varied markedly with season. In spring when discharge was high mesohabitats showed considerable overlap on the basis of their faunal communities. As the year progressed and discharge declined so mesohabitats became more distinct. There is a continual expansion and contraction of available mesohabitat. Thus instream macrophyte beds increase at the expense of open water habitat, and marginal vegetation encroaches into the main flow. The stream therefore consists of a dynamic system of mesohabitats whose borders are best defined in low flow periods. The 'distinctness' of the mesohabitats meant that it was possible to determine their proportion along a 300 m reach of the Mill Stream. The application of these data is discussed. This information together with data on the 'quality' (species richness, faunal abundance, rarity) can be used to categorise a stream at reach level and assess the effect of management (flow changes, dredging, bank clearance and weed-cutting) on the distribution of mesohabitats and hence on the faunal assemblages associated with them. This could be considered a more useful scale than point sampling when considering fish and macrophytes as well as invertebrates.