Many data in the literature suggest that patients on peritoneal dialysis (PD) show a gradual loss of peritoneal membrane function, leading to the failure of PD.1 Peritoneal membrane dysfunction is probably due to enhanced dissipation of the glucose-dependent osmotic gradient across the peritoneal membrane and loss of ultrafiltration capacity.2 One of the well-known causes of the deterioration of peritoneal membrane function is the cyto-toxicity of PD fluids, all containing glucose; its presence in the peritoneal solution as osmotic agent leads not only to the formation of glucose degradation products3,4 but also of reactive carbonyl compounds, such as glyoxal,methylglyoxal and 3-deoxyglucosone, deriving from heat-sterilization of PD fluids.5,6 These compounds lead to ‘carbonyl stress’, and play a pivotal role in peritoneal membrane dysfunction, not only modifying peritoneal matrix proteins and altering their structure but also reacting with mesothelial and endothelial surface proteins and, finally, initiating a range of inflammatory responses.3,7,8 In PD, reactive dicarbonyl compounds, precursors of advanced glycation end-products (AGEs), may derive from two sources: the peritoneal solution and the circulation, due to the condition of carbonyl stress present in uremia.9 AGEs are typical of diabetes.10 They accumulate on serum and tissue proteins in diabetic subjects, and their role in the pathogenesis of long-termdiabetic complications has been proved.11–13 However, recent studies have also shown that AGEs accumulate in the serum and tissues of patients with chronic renal failure, independently of the presence of diabetes.12,14 ‘Carbonyl stress’ has been invoked to explain the long-term complications associated with chronic renal failure and dialysis, such as atherosclerosis and dialysis-related amyloidosis.15 In this panorama, the development of specific analyticalmethods capable of evaluating glyoxal and methylglyoxal levels either in the PD solutions or in the dialysis fluids are certainly of great interest, allowing physicians to establish the role of carbonyl stress in the general pathological picture and to undertake possible therapies. Thirty-six 600-l plasma samples – 20 from healthy subjects and 16 from uremic patients before and after 12 h of peritoneal dialysis – were analyzed. In order to deproteinize plasma and dialysate samples, the approach employed by Tsukushi et al.22 was used, which is based on ultrafiltration on a Centricon membrane. Glyoxal and methylglyoxal levels were expressed as g/l. Values were expressed as means+- SD (standard deviation) or means +- SEM (standard error of mean). To compare mean values between quantitative variables, Student’s t-test for paired and unpaired data was applied. A statistically significant difference was accepted at p