Bivalve molluscs can concentrate contaminants, including pathogenic microorganisms, from the water column during their normal filter-feeding activity. In the European Union, the risk of human and animal faecal contamination in bivalves is estimated by determining the concentration of Escherichia coli in time-series samples from production areas. A structured field study was undertaken to determine the extent to which such concentrations varied between sites, sampling occasions and shellfish species and to determine the residual variability of the method. E. coli was enumerated in three species of bivalve mollusc (Crassostrea gigas, Mytilus spp. and Pecten maximus) co-located in each of three geographically separate commercial shellfisheries. The data were subjected to analysis of variance (ANOVA). This showed that the effects of site, sampling occasion, species and site/sampling occasion interaction were all significant. The proportion of variation due to site was markedly greater than that due to other factors. Post-ANOVA analysis showed that the concentration of E. coli in P. maximus was significantly higher than in the other two species. Mytilus spp. and C. gigas exhibited comparable levels of E. coli. The observed standard deviation of the most probable number method in the study was 0.33 log10.