Peritonitis (P) is the most important infectious complication in peritoneal dialysis (PD), but disagreement remains about the impact of PD modality on the frequency of P episodes. We compared indices linked to P between patients on continuous ambulatory PD (CAPD) and those on automated PD (APD) in the short and very long term. The study included 9 prevalent and 72 incident patients on CAPD (twin-bag system) and 37 incident patients on APD from 4 August 1995 to 31 March 2011. Data were collected prospectively from our database. The cumulative P rate (CPR) by weighted Student t-test (p < 0.05) expressed as the probabilities of remaining free of a first P episode and remaining free of all P episodes were compared by the Kaplan-Meier method (log rank p < 0.05). The proportion of patients with P (PPP) per group (chi2 p < 0.05) and the relative risk (RR) of P were also calculated. The CAPD patients included 51 women and 30 men [mean age.: 50.2 +/- 14.21 years (range: 15-82 years); duration of treatment: 3579 patient-months (mean: 44.2 +/- 34.2 patient-months); P episodes: 111] had a CPR of 0.38 episodes (1 episode in 31.35 patient-months) and a P rate per year (PRY) in the range 0.08-0.89 episodes. The APD patients [23 women and 14 men; mean age: 53.62 +/- 13.61 years (range: 26-78 years); duration of treatment: 1718 patient-months (mean: 46.4-38.3 patient-months); P episodes: 46] had a CPR of 0.32 episodes (1 episode in 373 patient--months) and a PRY in the range 0-1.12 episodes. At 1, 3, 5, and 10 years respectively, a first P episode occurred in 73%, 36%, 23%, and 8% of CAPD and 72%, 59%, 35%, and 23% ofAPD patients (log-rank p = 0. 056; CPR: p = 0.01; PPP: chi2 p = 0.39), for a RR of 0.83 linked to CAPD compared with APD. A lower P frequency and a trend of less time to first P was observed for APD compared with CAPD. The CAPD patients did not show a higher risk of developing P than did the APD patients. For comparisons between the modalities, CPR is a reliable index in the very long term at risk. Depending on the size of the population and the time at risk, the PRYobtained on short follow-up could result in a misinterpretation of the performance of each treatment modality.