In the interest of studying the prevention of chronic peritoneal dialysis infections, serial studies of the bacterial epidemiology in peritonitis and of antibiotic prophylaxis, respectively, were carried out. For 18 months, prospective evaluation of catheter exist site cultures, performed at the time patients developed acute peritonitis, showed that Staphylococcus aureus peritonitis was associated with concordant S. aureus at the exist site in 85% of cases, significantly more frequent than that for other organisms (P less than 0.02). Furthermore, active inflammation along with concordant culture results at the exit site characterized more than 60% of S. aureus peritonitis cases, also significantly more than that for other organisms (P less than 0.01). Over the ensuing 2 yr, patients beginning chronic peritoneal dialysis with a new percutaneously placed catheter were prospectively entered into a randomized, controlled trial of long-term antibiotic prophylaxis with trimethoprim-sulfamethoxasole. Patients receiving prophylaxis tended to have fewer episodes of peritonitis; however, the lower rate of peritonitis reached statistical significance only comparing patients who were S. aureus carriers at entry into the study to patients who were not S. aureus carriers. In particular, the prophylaxis trial seemed to reduce the specific incidence of S. aureus peritonitis overall, with S. aureus appearing in only 2 of 28 total peritonitis episodes among treated patients as compared with 11 of 37 total episodes among non-treated patients (P less than 0.01). Further analysis of the time to first peritonitis suggests that the effect of prophylaxis was most prominent during the first 3 months of therapy (P less than 0.02) rather than later in the course of treatment.(ABSTRACT TRUNCATED AT 250 WORDS)