We have investigated the febrile responses of New Zealand White rabbits to a Gram-negative pyrogen (bacterial lipopolysaccharide (LPS) from Salmonella typhosa), commonly associated with systemic infection, and a Gram-positive pyrogen (Staphylococcus aureus), more frequently associated with superficial soft tissue infection, each administered via one of four different routes (intravenous, intramuscular, subcutaneous or intraperitoneal) at each of three different doses (LPS: 0.1, 1 and 10 µg kg-1; S. aureus: 1.5 × 107, 1.5 × 108 and 1.5 × 109 cell walls kg-1). Intravenous administration of LPS evoked rapid, dose-dependent biphasic fever. Injection of LPS by the other routes also evoked dose-dependent fever. However, these fevers were monophasic, had increased latency of onset, and were of lower amplitude. It is important to note that a dose of approximately 10 and 100 times that of the standard intravenous dose was required to produce a similar peak rise in temperature when administered subcutaneously and intraperitoneally, respectively. Intravenous injection of the highest dose of S. aureus evoked dose-dependent biphasic fever, with short latency of onset, which was very similar to that induced by intravenous LPS. At lower doses, intravenous S. aureus induced monophasic fever. No fever occurred when the same doses of S. aureus were administered by any other route. We conclude that any of the four routes may be used for the study of LPS-induced fever, provided that the doses are adjusted. However, studies of S. aureus-induced fever, and detection of contamination with either pyrogen, requires intravenous injection. Experimental Physiology (2002) 87.3, 391-399.