Incubation of mouse serum albumin with the food borne carcinogen [2-14C]-Amino-3,8,-dimethylimidazo[4,5-f]quinoxaline (14C-MeIQx) in the presence of mouse hepatic microsomes and an NADPH-regenerating system in vitro resulted in the formation of adducts of MeIQx with albumin, which increased proportionately with time for at least 120 min (approximately 1 pmol equivalents/mg of protein/min). We have previously shown in male Swiss Webster mice in vivo that 14C-MeIQx bound covalently to serum proteins and that the formation of adducts was dose dependent. 14C-MeIQx (100 mg/kg, i.p.) was administered to male (MF1) mice which were killed 24 h later. Serum albumin was purified by affinity chromatography and covalent binding of 14C-MeIQx was assessed. Total covalent binding of MeIQx to albumin was 14.0 +/- 5.2 pmol per mg albumin, which was 5-fold greater than to haemoglobin. Following mild acid hydrolysis, 1.25 pmol MeIQx per mg albumin was liberated as free amine, as determined by gas chromatography negative ion mass spectrometry (GC-MS). This represents 9% of total MeIQx adducted to albumin in vivo (cf 1.3% adducted to haemoglobin). These results suggested that adducts of MeIQx with serum albumin should provide a significantly more sensitive dosimeter than those with haemoglobin. We therefore investigated this approach with serum protein samples from three volunteers. Human serum albumin and non-serum albumin protein fractions were separated by affinity chromatography, before being subjected to GC-MS analysis for hydrolysable adducts of MeIQx. The levels of MeIQx in control samples, and from the release of the putative sulphinamide adducts in hydrolysed samples were below the limits of detection of the GC-MS assay (29 +/- 2.6 amol MeIQx/mg albumin). Despite an increase of 2 orders of magnitude in sensitivity, compared with haemoglobin, it is unlikely that the sulphinamide adduct of MeIQx with human serum albumin can be used as a dosimeter for human aminoimidazoazaarene exposure.