An epizootic of subclinical and clinical mastitis caused by Serratia marcescens was investigated in a 1,000-cow dairy farm in California. Serratia marcescens was isolated from 13 to 18% of composite milk samples obtained from lactating dairy cows. During monthly milk sampling performed during a 4-month period, S marcescens was isolated from 38.8 to 62.3% of composite milk samples obtained from cows from which S marcescens was previously isolated. Few cows infected with S marcescens had evidence of clinical mastitis. Somatic cell count value was associated with isolation of S marcescens. Cows with somatic cell counts greater than 500,000 were 5.48 times as likely to have intramammary infections with S marcescens, compared with cows with somatic cell count less than or equal to 500,000. Lactation number also was associated with S marcescens intramammary infection. After adjusting for the effect of lactation number, cows with high somatic cell count values were 2.98 times as likely to have intramammary infection with S marcescens, compared with cows with low somatic cell counts. Infection with S marcescens was independent of days in lactation, production string, and daily milk production. Eleven months after the beginning of the epizootic, S marcescens was isolated from organic bedding samples obtained from the dairy. Despite numerous attempts, other sources of S marcescens could not be identified on this dairy.