A variety of methods have been developed for the measurement of tumor necrosis factor (TNF)-alpha synthesis by immune cells. Here we have compared the results of the most common used methods, including in vitro stimulation of whole blood or peripheral blood mononuclear cell (PBMC) cultures with phytohaemagglutinin (PHA) or lipopolysaccharide (LPS) and RT-PCR analysis of TNF-alpha transcription in unstimulated PBMC. When we used EDTA treated blood samples we observed a significant correlation between the PHA and LPS stimulated TNF-alpha responses in whole blood or PBMC cultures. In contrast, TNF-alpha concentrations obtained from PHA and LPS stimulated whole blood cultures from citrate-treated blood did not show a correlation. We also found that the PHA stimulated TNF-alpha response was significantly higher in PBMC than in whole blood cultures, whereas the highest LPS stimulated TNF-alpha response was observed in citrate-treated blood. Moreover, the TNF-alpha response in both, citrate and EDTA treated whole blood cultures was significantly higher after LPS than after PHA stimulation. In contrast, in PBMC cultures the PHA stimulated TNF-alpha response was significantly higher than the LPS stimulated response. The results of RT-PCR analysis revealed a significant correlation with the PHA stimulated TNF-alpha response, both in whole blood assays and in PBMC cultures. In addition our results demonstrate that these different methods can only be compared when the influence of external factors such as the immediate processing of blood samples or the use of an appropriate anticoagulant and stimulant is considered.