We reviewed all episodes of peritonitis associated with exit site and/or tunnel infection (n = 87; rate, 0.1/yr; 13% of all peritonitis episodes) occurring from 1979 to 1995. The exit site or tunnel infection was diagnosed at the time or shortly after the patient presented with peritonitis in 66% of the episodes. In the other one third the exit site or tunnel infection was diagnosed a median of 40 days prior to the development of peritonitis. Staphylococcus aureus accounted for 52% of episodes. Pseudomonas aeruginosa was the next most common organism. In 63 (72%) of the episodes the catheter was removed to resolve the infection at a median of 8 days (range, 0 to 226 days) from the onset of peritonitis. Catheter removal after 5 days predominately for refractory peritonitis (n = 23; median time to removal, 8 days) or relapsing peritonitis (n = 11; median time to catheter removal, 103 days). Patients with relapsing peritonitis suffered two to four episodes prior to removal of the catheter. Patients with peritonitis associated with tunnel infection were more likely to lose their catheter than patients with peritonitis associated with exit site infection (86% v 58%), while Staphylococcus epidermidis infections were less likely to result in catheter loss compared with all other organisms (15% v 82%). After a protocol to reduce S aureus catheter infections was implemented in 1990, the rate of catheter-related peritonitis decreased from 0.14/yr to 0.05/yr due to a decrease in S aureus episodes. We conclude that peritonitis episodes associated with a tunnel infection infrequently resolve without catheter removal. Delayed catheter removal in such circumstances often results in refractory or relapsing peritonitis. Therefore, catheter removal should be done promptly. Antibiotic prophylaxis for S aureus can reduce catheter-related peritonitis.