This study investigated the effect of in vitro exposure to caffeine, and its major metabolite paraxanthine, at concentrations relevant to typical caffeine consumption in humans, on lipopolysaccharide (LPS)-stimulated cytokine production in human whole blood. In addition, a role for the cyclic AMP/protein kinase A (PKA) pathway in the immunomodulatory effect of caffeine was investigated. Diluted whole blood (taken following >/=15 h abstinence from caffeine-containing food and beverages) was preincubated with caffeine or paraxanthine (10-100 microM) and stimulated with LPS (1 proportional, variant g/ml) for 24 h. The proinflammatory cytokines tumour necrosis factor (TNF)-alpha, interleukin (IL)-1beta and IL-12, and the antiinflammatory cytokine IL-10 were measured in cell-free supernatants. Whilst caffeine and paraxanthine had little or no effect on IL-10, IL-1beta, or IL-12 production, TNF-alpha production was suppressed in all individuals studied. The effect was statistically significant at 100 microM and consistent across seven experiments performed. Although not statistically significant, a similar effect was observed with paraxanthine. Caffeine (100 microM) also increased intracellular cyclic AMP concentrations in LPS-stimulated monocytes isolated from whole blood. Moreover, the effect of caffeine on TNF-alpha production was abolished by pretreatment with the protein kinase A inhibitor Rp-8-Br-cAMPS (10(-4) and 10(-5)M). To conclude, this study demonstrates that concentrations of caffeine that are relevant to human consumption consistently suppress production of the proinflammatory cytokine TNF-alpha in human blood and that this effect is mediated by the cyclic AMP/protein kinase A pathway.