As part of the EU-funded project "CLUSTER" a database was constructed of herring schools identified during a series of acoustic surveys in the northwest North Sea. Among other descriptors, the database included each schools' height, length, and acoustic backscattering energy (S,,). The number of schools per nautical mile. EDSU (Elementary Distance Sampling Unit) was also recorded. The relationships between local school count and school backscattering energy to time-of-day and location were first modelled using multiple regression techniques. The results indicate a considerable degree of non-linear dependency on both time-of-day and location. Herring-school counts per EDSU tended to be high during the middle part of the day and lower at dawn and dusk and were higher along the continental shelf edge about 130 m west of Orkney and Shetland. The regression models, by definition, also allow variability due to each explanatory variable to be assayed and divided. This feature meant that their output could be used to explore further into the relationships among the schools. In this paper the residual variability from the regression models is used to describe density-dependent relationships among herring schools, i.e. we asked "To what extent does local herring school abundance influence the size (backscattering energy) of a given school?" It is concluded that herring school size is regulated mainly by location and time-of-day and that "measured" school size is not influenced by the local "school count per EDSU". The results and their implications are discussed. (C) 2002 International Council for the Exploration of the Sea. Published by Elsevier Science Ltd. All rights reserved.