The hospital records of all patients on continuous ambulatory peritoneal dialysis (CAPD) who were followed from January 1992 to June 1993 in a third level pediatric hospital were retrospectively reviewed to compare the peritonitis rates of groups of patients according to the dialysis system employed. All patients had Tenckhoff catheters installed by open technique. The groups were A) traditional spike connection system (37 patients), and B) disconnect systems (16 patients) that comprised the O-set system (10 patients) and the basic Y-set system (6 patients). Age was 11.9 +/- 3.7 years (mean +/- SD, range 3 months to 17 years). Follow-up time was 8.5 +/- 4.6 months for the traditional system group (304 cumulated months) and 7.2 +/- 3.6 months for the disconnect systems group (116 cumulated months). Peritonitis rates were 1 episode every 10 months for group A, versus 1 episode every 58 months for group B (p < 0.05). Our results strongly suggest that disconnect systems for peritoneal dialysis can significantly decrease peritonitis rates, although longer follow-up periods are needed.