The relationship between the abundance of taxa and life forms of lithorheophile macrobenthos and its variability were studied based on 200 quantitative samples from six territories of the Palaearctic (Moscow province, northwestern Caucasus, eastern Carpathians, northern Karelia, South Urals, and Altai mountains). The set of taxa predominant in the communities and their ecology are described. It is found that community structure varies strongly, depending on the characteristics of each region, on the size of the watercourse, and on the season. Six types of biocenoses are recognized by means of the Braun-Blanquet method, each characterized by its peculiar set of predominant life forms and families rather similar in different territories. The differences between these types are related to the size and the hydrological conditions of the watercourse. Biocenosis 1 is typical to smal brooks (up to 0.01-0.1 m3/s), characterised by the predominance of detritophagous animals non-specific to the type of food (Gammarus, Nemoura, Limnephilidae). In biocenosis 2a (large brooks with water flow 0.03-0.3 m3/s and velocity 0.1-0.3 m/s), almost immobile shell scrapers (Ancylus, Silo, Agapetes, Glossosoma) are predominant. Biocenosis 2b (large brooks with velocity 0.3-0.5 m/s) have a more or less balanced set of fundamental lithorheophile life forms. Biocenosis 2c (large mountain brooks with velocity 0.5-1 m/s) is characterised by specialized scrapers of the rapids (Epeorus and Diomesa) and filterers (Simuliidae). In biocenosis 3 (small rivers), sedentary filterers (Hydropsychidae, Simulliidae) are predominant; scrapers also play a significant role. Biocenosis 4 (rivers with water flow more than 3 m3/s, thick incrustations, and silted stones on the bottom) has predominant filterers (Hydropsychidae) and vermiform algophagous animals inside the incrustations (Orthocladius, Psychomyia). Significant variability in community structure unrelated to the environmental factors is revealed within each of the biocenotic types. It is concluded that the quantitative structure of lithorheophile communities cannot be reliably predicted at the present level of knowledge, and the extrapolation of data is possible only at the level of species composition and ranked distribution of abundance.