The elaboration of a sensitive bioassay for assessment of tumour necrosis factor alpha (TNF-α) in a defined medium is described. The assay is based on the cytotoxic effect of TNF-α on a target cell line, the murine fibrosarcoma WEHI 164 clone 13. Cytotoxicity was assessed by detecting the rate of tetrazolium salt reduction employing a spectrophotometer (ELISA-reader). A similar bioassay was used previously to assess TNF-α, though this was dependent on cell growth in a medium containing serum. By employing a synthetic serum replacement, the WEHI cells were adapted to growth in a defined medium which allowed both the propagation of the cell line and the assay to be performed under completely defined conditions. Thus, factors in serum that may influence the TNF-α assessment, such as growth factors, cytokines, soluble cytokine-receptors and macroglobulin, were avoided. The only protein required in this bioassay was insulin, while albumin was added as a carrier protein and to protect the cytokine against loss of biological activity during multiple freeze and thaw cycles. The present assay was optimised to achieve a high sensitivity and, by testing endogenous TNF-α originating from the macrophage-like cell line RAW in both the serum-free and serum-based assay, we found the highest sensitivity in the assay based on defined medium. The LC50 of recombinant mouse and human TNF-α were in the serum-free and serum-based assays considered to be 25 and 50 pg mL-1, respectively. The demonstration of a culture condition that enables long-term cultivation of target cells and a bioassay in a completely defined medium is in our opinion a substantial contribution to more reliable cytokine assessment.