The effects of nimesulide and diclofenac on lipopolysaccharide (LPS)-induced rectal temperature changes and serum tumour necrosis factor (TNF)-alpha elevation were investigated in rats. LPS (Escherichia coli O111:B4; 50 microg/kg, intraperitoneally) produces a dual body temperature response, in which initial hypothermia precedes fever. Serum TNF-alpha levels rise during the initial phase of the induced hypothermia. Nimesulide, a preferential inhibitor of cyclooxygenase-2 (0.05, 0.5 or 1 mg/kg, subcutaneously) completely abolished the hypothermia, resulting in an acceleration of the fever phase. However, the peak and plateau phases of fever were not changed by nimesulide treatment. Nimesulide (0.5 mg/kg) partially prevented serum TNF-alpha elevation. The non-selective cyclooxygenase inhibitor diclofenac inhibited hypothermia at all doses tested (0.03, 0.3 or 3 mg/kg, subcutaneously) although fever was completely abolished at the 3 mg/kg dose only. Diclofenac also partially abolished the elevation in serum TNF-alpha levels, but at the highest dose only (3 mg/kg). These data suggest that nimesulide and diclofenac can preferentially inhibit LPS-induced hypothermia at doses that do not abolish fever in rats. Both these drugs also reduced elevated TNF-alpha levels, a fact which may, at least partly, explain the antihypothermic effect of nimesulide.