1. A survey of post-operative wound infection was done in 1959–60 on 559 surgical patients admitted to a provincial general hospital.2. Clinical evidence of post-operative wound sepsis was observed in 71(12·7 %), suppuration in 51(9·1 %) and staphyloccocal wound sepsis in 48 (8·6 %).3. Seventeen of the patients died in hospital. Although 5 of these had septic wounds, the sepsis did not appear to have been the cause of death.4. Contrary to some reported findings, the post-operative wound sepsis rate was considerably lower in patients who were staphylococcal nasal carriers on admission to hospital than in non-carriers. Nine out of 153 carriers (5·9 %) developed wound sepsis and 36 out of 385 non-carriers (9·4 %). When allowance is made for 3 highly probable self-infections, the incidence of wound cross-infection was 3 % in carriers and 9 % in non-carriers.5. The excess of sepsis in non-carriers could not be explained in terms of different age or sex distribution in carrier and non-carrier groups, nor by differences in the types of operation undergone or in degree of exposure to staphylococcal contamination.6. The excess of sepsis in non-carriers was accounted for by the patients whose wounds had drains rather than by clean-stitched wounds.7. Those carriers who harboured a staphylococcus of the 80/81 group in the nose on admission to hospital had a higher incidence of wound sepsis than carriers of other phage types or staphylococcus.8. Three probable instances of wound self-infection occurred, and in the early stages of the survey at least 5 wound infections were probably directly caused by two members of the theatre staff carrying staphylococci of the 80/81 group in the nose and with existing skin sepsis or a recent history.9. The survey findings and a study of the literature suggested: (a) that the acquisition of a nasal staphylococcus in hospital was probably, as a rule, evidence of exposure to staphylococeal contamination and not a determinant of wound sepsis, unless the patient also became a skin carrier; (b) that a small proportion of patients are self-infected, some are directly infected by theatre personnel, and the wounds of other patients are directly or indirectly contaminated by staphylococci from the ward environment; (c) that wool or cotton fluff contaminated with staphyloccoci may cause wound sepsis by falling into open wounds and as foreign bodies induce a significant reduction in the minimum pus-forming dose of staphylococci; (d) that before ascribing high or low sepsis rates to factors such as the nasal carrier state, the relevant groups should be shown not to differ materially in respect of other factors known to influence the incidence of wound sepsis.We gratefully acknowledge the help and encouragement of our surgical colleagues Mr H. P. Guerrier, Mr P. J. W. Monks, Mr J. Macpherson and Mr C. C. Jeffery; and of Dr C. P. Warren, consultant pathologist to the Torbay Hospital, Mrs M. Stamp, Matron, Sister E. M. Cottrell, first Infection Control Sister at the Torbay Hospital, and her successor Mr H. Street.Acknowledgement is also due to the South Western Regional Hospital Board for its encouragement and for a research grant for technical aid, without which the surveys described would not have been possible.