A series of least-squares analytical models were developed in a cross-sectional epidemiologic study of the observed variability in within-herd high somatic cell count (SCC) prevalence, a measure of mastitis prevalence in dairy herds. The dependent variable, high SCC prevalence, was calculated as the 12-month rolling herd average percentage of lactating cows with milk SCC in excess of 283 000 cells ml-1. The first analysis involved the results of a bacteriologic survey of bulk-tank milk samples for the presence of Streptococcus agalactiae and coagulase-positive staphylococci, the two most common contagious intramammary pathogens. The presence of either pathogen in bulk-tank samples was associated with significantly higher high SCC prevalence. The second analysis involved responses to a questionnaire concerning management and mastitis control practices. Both the practices of post-milking teat dopping and dry-cow antibiotic therapy were associated with significantly lower high SCC prevalence. The third analysis combined the data collected for the first two analyses so that the independent variables included both bulk-tank bacteriologic results and management and mastitis control practices. This model was able to explain a greater proportion of the variability in high SCC prevalence than either of the other two models. There were three variables associated with significant decreases in high SCC prevalence namely the absence of Streptococcus agalactiae in the bulk tank milk, the adoption of post-milking teat dipping and the practice of dry-cow antibiotic therapy of all cows. Milk somatic cell counting is now widely accepted and practiced in many countries, and individual-cow SCC data are available from large numbers of herds at a minimal expense. By corroborating the role of post-milking teat dipping and dry-cow antibiotic therapy in mastitis control programs, this study establishes the usefulness of high SCC prevalence data for epidemiologic studies of mastitis control practices.