BACKGROUND The advanced stage of the Maillard reaction, which leads to the formation of advanced glycation end products (AGE), plays an important role in the pathogenesis of angiopathy in diabetic patients and in the aging process. N(epsilon)-(carboxymethyl)lysine (CML) is thought to be an important epitope for many of currently available AGE antibodies. However, recent findings have indicated that a major source of CML may be by pathways other than glycation. A distinction between CML and non-CML AGE may increase our understanding of AGE formation in vivo. In the present study, we prepared antibodies directed against CML and non-CML AGE. METHODS AGE-rabbit serum albumin prepared by 4, 8, and 12 weeks of incubation with glucose was used to immunize rabbits, and a high-titer AGE-specific antiserum was obtained without affinity for the carrier protein. To separate CML and non-CML AGE antibodies, the anti-AGE antiserum was subjected to affinity chromatography on a column coupled with AGE-BSA and CML-BSA. Two different antibodies were obtained, one reacting specifically with CML and the other reacting with non-CML AGE. Circulating levels of CML and non-CML AGE were measured in 66 type 2 diabetic patients without uremia by means of the competitive ELISA. Size distribution and clearance by hemodialysis detected by non-CML AGE and CML were assessed in serum from diabetic patients on hemodialysis. RESULTS The serum non-CML AGE level in type 2 diabetic patients was significantly correlated with the mean fasting blood glucose level over the previous 2 months (r = 0.498, p < 0.0001) or the previous 1 month (r = 0.446, p = 0. 0002) and with HbA(1c) (r = 0.375, p = 0.0019), but the CML AGE level was not correlated with these clinical parameters. The CML and non-CML AGE were detected as four peaks with apparent molecular weights of 200, 65, 1.15, and 0.85 kD. The hemodialysis treatment did not affect the high-molecular-weight protein fractions. Although the low-molecular-weight peptide fractions (absorbance at 280 nm and fluorescence) were decreased by hemodialysis, there was no difference before and after dialysis in the non-CML AGE- and CML-peptide fractions (1.15 and 0.85 kD fractions). CONCLUSIONS We propose that both CML and non-CML AGE are present in the blood and that non-CML AGE rather than CML AGE should be more closely evaluated when investigating the pathophysiology of AGE-related diseases.