Glucose-containing peritoneal dialysates are used to generate an osmotic gradient for the convective removal of water and sodium. Predictive equations were developed to estimate glucose absorption without having to formally measure changes in dialysate glucose. In view of the changes in peritoneal dialysis prescriptions over time, we compared predicted and measured glucose absorption. We measured peritoneal glucose losses when peritoneal dialysis patients attended their first assessment of peritoneal membrane function, and compared this to glucose exposure and Kidney Disease Outcomes Quality Initiative, Grodstein and Bodnar predictive equations. We studied 689 patients; 329 (56.9%) males, 53 (37.1%) diabetics, with mean age 57.1 ± 16.2 years, with 186 treated by automated peritoneal dialysis cyclers and 377 by automated peritoneal dialysis with a daytime icodextrin exchange and 126 by continuous ambulatory peritoneal dialysis. Using Bland -Altman analysis, all equations demonstrated systematic bias overestimating glucose absorption with increasing glucose absorption. For continuous ambulatory peritoneal dialysis patients, the Kidney Disease Outcomes Quality Initiative formula underestimated glucose absorption (bias 188 (-39 to 437) mmol/day, as did Grodstein (bias 37.9 (-105 to 29) mmol/day, whereas mean bias for Bodnar was -29 (-130 to 180)). There was systematic overestimation for all equations for both automated peritoneal dialysis with and without a daytime exchange, with increasing bias with greater glucose absorption. Although formally measuring peritoneal glucose absorption is time consuming and requires patient co-operation, current predictive equations overestimate glucose absorption and do not provide accurate estimations of glucose absorption particularly for automated peritoneal dialysis patients.