Advanced glycation endproducts (AGEs) have been implicated in the pathophysiology of coronary heart disease in ageing, diabetes and renal disease. Competitive enzyme-linked immunosorbent assays (ELISAs) have been developed to measure these compounds in serum, but as recognition of AGEs is both carrier protein- and antibody-dependent standardisation is problematic. We report here on another barrier to standardization, as yet unrecognised. During the development of an AGE ELISA, we found that serum samples did not dilute in parallel to AGE standards or each other. This finding was confirmed by recovery studies that showed over-recovery of AGEs at high serum concentrations, but under-recovery at high dilutions of serum in assay buffer. We developed an inhibition assay to detect factors in serum capable of interacting directly with AGEs immobilised on microtitre plates. Binding of these factors prevented recognition of AGEs by a CML monoclonal antibody and a polyclonal anti-AGE antibody, and was neither sugar- nor carrier protein-dependent. We detected the presence of this factor in all human sera tested and also in foetal calf serum. Pre-incubation of sera with AGEs or heat-treatment at 56 degrees C for 30 min. significantly reduced this binding. We are currently investigating the nature of this factor and the possibility that it may be complement. The effect of this factor on immunoassays for AGEs can only be detected by performing parallelism and recovery studies and we suggest the use of the method referred to in this paper to aid interpretation of parallelism data.