The purpose of the study was to determine the epidemiological relationships in three unrelated cases of neonatal late-onset Group B streptococcal (GBS) disease and maternal breast-milk infection with GBS. All deliveries were by cesarean section; case 1 was at term, and cases 2 and 3 were at 32- and 33-wk gestation, respectively. Case 1 relates to a mother with clinical mastitis and recurrent GBS infection in a 20-day-old male infant. Following antibiotic therapy and cessation of breast-feeding, the infant recovered without sequelae. Case 2 refers to a mother with clinical mastitis and the occurrence of late-onset GBS disease in 5-wk-old male twins. Despite intervention, one infant died and the second became ill. Following antibiotic therapy and cessation of breast-feeding, the surviving infant recovered without sequelae. Case 3 refers to a mother with sub-clinical mastitis and late-onset GBS infection occurring in a 6-day-old female twin. Following intervention, the infant recovered but suffered a bilateral thalamic infarction resulting in developmental delay and a severe seizure disorder. Following recovery of GBS from an inapparent mastitis and cessation of breast-feeding, the second infant remained well. Blood cultures from all affected infants and maternal breast milk were positive for GBS. Epidemiological relationships between neonatal- and maternal-derived GBS isolates were confirmed by a random amplified polymorphic DNA polymerase chain reaction assay (RAPD-PCR). This study is significant in that it has demonstrated that maternal milk (in cases of either clinical or sub-clinical mastitis) can be a potential source of infection resulting in either late-onset or recurrent neonatal GBS disease.