In conscious, freely moving guinea pigs, tumor necrosis factor (TNF)-alpha and TNF-beta, infused into the aortic arch within a period of 45 min at a dosage of 5 micrograms/kg, induced different thermal responses. TNF-alpha evoked a biphasic elevation of abdominal temperature, both phases together lasting longer than 6 h. In response to infusions of TNF-beta, the first phase, lasting approximately 120 min, was the same as was observed in response to TNF-alpha, whereas the longer second phase of temperature increase was missing. When the infusion of TNF-alpha was repeated four times at intervals of 3 days, the second phase of the increase in abdominal temperature (120-360 min after start of infusion) tended to decrease in response to the third and was significantly attenuated in response to the fourth infusion of TNF-alpha. A control group of guinea pigs received four infusions of solvent (0.9% sterile pyrogen-free saline). Another 3 days after the fourth infusion of TNF-alpha or solvent, all animals were injected with 20 micrograms/kg bacterial lipopolysaccharide (LPS from Escherichia coli; intramuscular injection). In those guinea pigs having developed a reduced responsiveness to TNF-alpha, the first phase of LPS-induced fever was significantly suppressed, whereas the second phase tended to be enhanced, compared with animals having received four infusions of solvent. In conclusion, guinea pigs develop a reduced responsiveness to TNF-alpha after its repeated administration. In the state of lower reactivity to exogenous TNF-alpha, a reduced response of the first phase of LPS-induced fever (during which endogenous TNF-alpha is released) can be observed. This indicates that endogenous TNF-alpha may contribute to LPS-induced fever only in the initial phase of the febrile response of guinea pigs.