We report an outbreak of 23 neonatal group A streptococcal infections including two cases of septicaemia in a nursery for the newborn. At the same time, 19 mothers had puerperal endometritis. The outbreak lasted for about 2 months and could not be eradicated by ordinary hygienic measures. In a colonisation study, 19 of 90 umbilical stumps studied on the day of discharge from hospital harboured group A streptococci, all of the same epidemic strain T28 provisional M-type 2841/opacity factor positive (T28/MPT2841/OF+). Chlorhexidine gluconate was therefore applied daily to the umbilical stumps. The outbreak came to an end with the introduction of this procedure. The time interval between delivery and recognition of infection was shorter in infants than in mothers. There were also only two infected mother-infant pairs. Taken together, this suggests that infected umbilical stumps on symptomless infants were a likely source of the maternal infections. Throughout 1 year from the introduction of the chlorhexidine treatment, we have not seen a case of group A streptococcal infection in neonates. Surveillance cultures from umbilical cords after the outbreak have also been negative. We conclude that bacteriological surveillance of umbilical stumps is valuable in recognising an outbreak and should be seriously considered when a group A streptococcal infection has been recorded in a nursery.