We conducted a single-blind, prospective randomized study on the use of the Y-set disconnect system (Ultraset) (U) versus the conventional (C) spike system to assess the peritonitis rate, exit-site infection (ESI), clinical outcome, the resulting hospitalization rate, and recurrent costs. Forty new end-stage renal failure patients admitted to the dialysis program were recruited into the study and 20 each were randomly allocated to the U and C systems. They were studied for a period of 12 months. The mean number of days required to train patients for the U and C systems were 8.6 and 9.8 days, respectively. The peritonitis rates for the U and C systems were one episode every 17 and 11.4 patient-months, respectively. The ESI rates for the U and C systems were one episode every 26.4 and 21.6 patient-months, respectively. Four catheters were removed due to fungal peritonitis (three with the C system and one with the U system). As related to peritonitis, patients on the C system required 57 hospital-days while those on the U system required 28 days per year. On cost analysis, the extra cost required for the U system can be offset by the other expenses incurred for events related to more infections on the C system. It is concluded that for the similar cumulative costs required for the patients on the two systems, the Y-set disconnect has a better morbidity profile than the conventional spike system.