1. The aerobic faecal flora of healthy infants was found to comprise coliform organisms, streptococci and occasionally yeast and spore-bearing organisms. Coliforms and streptococci occurred more frequently in the stools of artificially fed than of breast-fed infants in whom micrococci occurred more frequently.2. Of 191 strains of streptococci isolated from healthy infants' faeces and tested serologically, 8·90% fell into Lancefield's group D. Of the stools examined 93·7% contained group D streptococci.3. A comparison of the group D streptococcal flora of the faeces of healthy infants and of infants associated with outbreaks of neonatal diarrhoea showed that the percentage of these organisms found in the latter was much higher than in healthy infants, the modes being 90–100 and 10–20% respectively.4. The same species and varieties of group D streptococci were found in both healthy and sick infants. Streptococcus faecalis was the predominant species.5. Serological typing of strains of group D streptococci from healthy infants from three maternity hospitals showed that particular types tended to predominate in each maternity unit and to recur over a considerable period of time, in each instance suggesting a reservoir of dissemination to successive infants as they arrived. With artificial feeding an increased diversity of serological types was found.6. Serological types of group D streptococci isolated from sick infants during ten outbreaks of neonatal diarrhoea showed a considèerable variety. Apart from one outbreak (A), where type I was isolated with great frequency, one or two types tended to predominate in each outbreak, but not to any great extent, and not always the same one.7. Types 1, 5, 8 and 9 were the types of group D streptococci most commonly found in both healthy and sick infants, but type 1 occurred in a significantly greater percentage of sick than of healthy infants.8. Tyrosine decarboxylase activities of group D streptococci were usually associated with serological types. The distribution of activities of these organisms was similar in healthy and sick infants.9. The possibility that neonatal diarrhoea may consist of more than one syndrome is discussed, and several of the outbreaks described here are classified into two groups.The author wishes to thank Dr A. T. R. Mattick for his constant interest in this work; Dr P. M. P. Shattock for advice and many valuable suggestions; Dr E. F. Gale for advice and material; Dr E. H. Gillespie, Dr A. M. McFarlan and Dr J. Wright for material and clinical data; Dr P. White and Miss Z. D. Hosking for assistance with the statistical computations; Mr L. G. Newland for technical assistance; and to express her appreciation to the Medical Research Council for a grant which enabled this work to be carried out.