BACKGROUND Time on treatment is associated with a greater risk of impaired ultrafiltration (UF) in peritoneal dialysis (PD) patients. In addition to increasing solute transport, a potentially treatable cause of impaired ultrafiltration, cross-sectional studies suggest that there is also reduced osmotic conductance of the membrane. If this were the case then it would be expected that the UF capacity for a given rate of solute transport would change with time. The purpose of this analysis was to establish how solute transport and UF capacity change relative to one another with time on therapy. METHODS Membrane function, using a standard peritoneal equilibration test, was measured at least annually in a well-characterized, single-center observational cohort of PD patients between 1990 and 2003. Demography included age, gender, original cause of renal failure, body surface area (BSA), validated comorbidity score, residual urine volume and urea clearances, peritoneal urea clearances, and plasma albumin. RESULTS Data from 574 new PD patients were available for analysis. Independent demographic factors associated with higher solute transport at baseline were male gender and higher residual urine volume. Throughout time on therapy there was a negative relationship between solute transport and UF capacity and a significant increase and decrease in these parameters, respectively. During the first 12 months of treatment, the increase in solute transport was not associated with the expected fall in UF capacity, a phenomenon that was not explained by informative censoring, but was associated with an increased, albeit weak, correlation with BSA. In contrast, later in treatment there was a disproportionate fall in UF capacity, more accelerated in patients developing UF failure. Early exposure to higher intraperitoneal glucose concentrations, in the context of more comorbidity and relative lack of residual renal function, was associated with more rapid deterioration in membrane function. CONCLUSIONS Despite a causal link between solute transport and UF capacity of the membrane, due to the effect of the former on the osmotic gradient, there is evidence of their longitudinal dissociation. This implies a change in the structure-function relationship with time on treatment that can, to some extent, be predicted from clinical factors present within the first year of treatment. Dialysis-induced membrane injury must involve at least two processes, for example, increased vascular surface area contact with dialysate combined with changes in hydraulic conductance due to scarring of the vessels and interstitium.