The peritoneal equilibration test (PET) with 3.86% glucose concentration (3.86%-PET) has been suggested to be more useful than the standard 2.27%-PET in peritoneal dialysis (PD), but no longitudinal data for 3.86%-PET are currently available. A total of 242 3.86%-PETs were performed in 95 incident PD patients, who underwent the first test during the first year of treatment and then once a year. The classical parameters of peritoneal transport, such as peritoneal ultrafiltration (UF), D/D(0), and D/P(Creat), were analyzed. In addition, the absolute dip of dialysate sodium concentration (DeltaD(Na)), as an expression of sodium sieving, was studied. D/D(0) was stable, and a progressive decrease in UF was observed after the second PET, whereas D/P(Creat) firstly increased and then stabilized. DeltaD(Na) was the only parameter showing a progressive decrease over time. On univariate analysis, D/D(0) and DeltaD(Na) were found to be significantly associated with the risk of developing UF failure (risk ratio (RR) 0.987 (0.973-0.999), P=0.04, and RR 0.768 (0.624-0.933), P=0.007, respectively), but on multivariate analysis only DeltaD(Na) showed an independent association with the risk of developing UF failure (RR 0.797 (0.649-0.965), P=0.020). UF, D/D(0), and D/P(Creat) changed only in those patients developing UF failure, reflecting increased membrane permeability, whereas DeltaD(Na) significantly decreased in all patients. The 3.86%-PET allows a more complete study of peritoneal membrane transport than the standard 2.27%-PET. DeltaD(Na) shows a constant and significant reduction over time and is the only factor independently predicting the risk of developing UF failure in PD patients.