OBJECTIVE To develop diagnostic and treatment strategies for peritoneal dialysis catheter exit-site and tunnel infections. METHODS All consenting peritoneal dialysis patients performing home dialysis through the University of Iowa Hospitals and Clinics Home Dialysis Training Center. This is a state-owned teaching hospital serving a rural population of approximately one million people in Iowa and western Illinois. METHODS Four dialysis nurses collected information on a prospectively designed data acquisition tool. Patients were randomly assigned to one of two treatment groups, intraperitoneal vancomycin plus oral rifampin or oral trimethoprim/sulfamethoxazole (TMP/SMX), and their initial antibiotic therapy determined by that assignment. If the infection was gram-negative, the initial antibiotics were discontinued and an alternative therapy begun. Therapy was initiated by the nursing staff and required physician notification within 48 hours. RESULTS There were 126 recorded catheter infections (exit-site, tunnel, or cuff infection) resulting in a rate of 0.67 episodes per patient year of exposure. Staphylococcus aureus was isolated from the majority (60%) of these events. Pseudomonas aeruginosa was the next most common isolate and accounted for 21% of infections. Rubor, dolor, and turgor are the classic signs of inflammation, and at least one of these was present in 79% of the episodes. Isolated pericatheter erythema or serous discharge was associated with a minimal risk (< 2%) of catheter loss. The presence of a purulent exit-site discharge identified patients who had a 30% chance of failing systemic antibiotic therapy and a 20% risk of catheter loss. The concurrent presence of exit-site tenderness or swelling identified the most severe infections. Staphylococcal infections responded equally well to local cleaning and vancomycin plus rifampin (86% cured) or oral trimethoprim/sulfamethoxazole (89% cured) therapy. Gram-negative infections were frequent (27%) and appeared to respond best to a combination of tobramycin and ciprofloxacin. CONCLUSIONS Exit-site/tunnel inflammation is detectable by patients and can be used to guide therapy. An isolated finding of erythema or serous discharge is not indicative of an acute infection and may not require systemic antibiotics. The presence of purulence identifies patients at risk for catheter loss, and these patients benefit from systemic therapy. The combination of a purulent exit-site discharge plus pericatheter tenderness or swelling identifies patients likely to suffer treatment failure and require subsequent catheter removal. The cure rate of gram-positive catheter infections treated with vancomycin plus rifampin was indistinguishable from that achieved with oral trimethoprim/sulfamethoxazole (p = 0.99).