It has been previously reported that different quantitative results can be obtained when TNF alpha is measured in biological fluids by bioassay and immunoassay. This is thought to be related to the presence of antigenic forms of TNF alpha that cannot be detected by bioassay, such as complexes with soluble receptors (sTNF-R) or TNF alpha monomers. In this work we have observed discrepancies between antigenic and bioactive TNF alpha even when we used a sandwich-ELISA, unable to detect TNF alpha monomers, based on antibodies that bind epitopes overlapping with the soluble-receptor binding site of TNF alpha. Moreover, we found that antigenic TNF alpha levels in the presence of p55-sTNF-R (sTNF-R1) measured by different immunoassays were variable, depending on the immunoreagents and incubation time. To investigate whether TNF alpha-soluble receptor complex dissociation occurring during assay incubations contributes to the variability of results, we studied the kinetics of TNF alpha-soluble receptor interactions and examined the effect of complex dissociation using different analytical systems. TNF alpha association (k(on)) and dissociation (koff) rate constants with sTNF-R1, measured by real-time biospecific interaction analysis, were 5.01 x 10(5) s-1 M-1 and 2.8 x 10(-4) s-1, corresponding to an equilibrium constant (Kd) of 0.59 nM and to a half life for these complexes of 38 min. Complex dissociation and differential changes in the TNF alpha-sTNF-R1 bound:free ratio, in different analytical systems, markedly affects TNF alpha quantification.