Tetrazolium salts like 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) or sodium 2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT) that form formazans after reduction are widely used to investigate cell viability. Besides cellular enzymes, some constituents of cell media and other substances reduce tetrazolium salts, thereby interfering with these assays. We describe here that different preparations of serum albumin from bovine or human origin can lead to a concentration-dependent increase in the signals of the XTT assay; therefore leading to an overestimation of cell numbers and to an underestimation of potential cytotoxic effects of compounds to be tested. The same effect was seen in the MTT assay with human serum albumin (HSA). We demonstrate that this reductive activity cannot be inactivated by proteolytic digestion, but that it is due to the free cysteine residue in albumin, and is also observed when cysteine or glutathione (GSH) are used. Binding of N-ethylmaleimide (NEM) to the free cysteine residue leads to a decrease of the albumin interference in the XTT assay.