BACKGROUND Dialysis catheters are a common cause of nosocomial septicaemia in haemodialysis units usually due to staphylococci, of which Staphylococcus aureus is the most pathogenic. In this study, the epidemiology and pathogenesis of dialysis catheter-related infections were studied, and methods to identify patients with these infections were evaluated. METHODS A one-year prospective study of 67 catheters in 43 haemodialysis patients was performed. Details about patients and catheters were obtained successively during the catheter period, and biochemical parameters expected to be related to infection were measured. After catheter insertion, all patients were screened for nasal carriage of S. aureus, and a culture was taken from the skin overlying the catheter insertion site. Once a week, cultures were taken from the insertion site and from the hub, and aerobic and anaerobic blood cultures were drawn from the catheter. If clinical signs of septicaemia occurred, peripheral blood cultures were also performed, when it was possible. RESULTS The incidence of septicaemia was 49% (21/43) in patients, and 56% of all cases were caused by S. aureus. The mortality was 14% (3/21) and the incidence of severe secondary complications to septicaemia was 24% (5/31). In all, 80% of all severe complications and 75% of all deaths from septicaemia were due to S. aureus. With respect to S. aureus septicaemia, the predictive values of positive (P) and negative (N) S. aureus cultures were as follows: nasal culture, P=36% (10/28), N=90% (35/39); culture from the insertion site, P=72% (13/18), N=98% (48/49); and culture from the hub, P=75% (3/4), N=83% (52/63). The risk ratio for S. aureus septicaemia was 26.2 (6.1-113), P=0.0001, according to the presence of S. aureus at the insertion site, and 3.3 (0.74-15.1), P=0.12 according to nasal carriage of S. aureus. The frequency of S. aureus phage-type Group 2 (43%) was much higher than the general frequency of this phage-type in Denmark, which is about 23%. Catheter blood cultures were positive although there were no clinical signs of septicaemia in 34% (23/67) of all catheter periods--84% of these were due to coagulase-negative staphylococci. CONCLUSIONS Dialysis catheter-related S. aureus septicaemia was highly unlikely if the patient had not been carrying S. aureus in the nose or at the insertion site during the time the catheter was in place. The best predictor of dialysis catheter-related S. aureus septicaemia was a positive S. aureus culture from the insertion site. Positive catheter blood cultures unrelated to any clinical signs of septicaemia occurred in one-third of all catheter periods, and 84% of these were due to coagulase-negative staphylococci.