When administered s.c. in the tail, both prostaglandin E2 (PGE2; 1.58-158.0 micrograms) and bradykinin (BK; 0.01-1.0 microgram) produced a dose-dependent allodynia in a warm-water tail-withdrawal assay in rhesus monkeys. PGE2 (A50 = 5.3 +/- 0.15 microgram) was 143-fold less potent than BK (A50 = 0.037 +/- 0.012 microgram) in producing allodynia at 42 degrees C. However, PGE2 (15.8 micrograms) was longer acting than an equieffective dose of BK (0.1 microgram), and the highest dose of PGE2 (158.0 micrograms) was the only treatment to produce allodynia when 38 degrees C water was used as the thermal stimulus, suggesting that PGE2 was a more efficacious allodynic agent than BK. Morphine (0.1-3.2 mg/kg) administered s.c. in the back completely blocked the allodynic effects of both BK (0.1 microgram) and PGE2 (15.8 micrograms), although morphine was more than twice as potent against BK (A50 = 0.26 +/- 0.085 mg/kg) than against PGE2 (A50 = 0.65 +/- 0.14 mg/kg). The effects of morphine were antagonized by the opioid antagonist quadazocine (0.1 mg/kg), indicating that morphine's effects were mediated by opioid receptors. The nonsteroidal anti-inflammatory drug ketorolac (0.32-10.0 mg/kg) administered s.c. in the back completely blocked the allodynic effects of BK (A50 = 0.60 +/- 0.095 mg/kg) but did not alter allodynia induced by PGE2. The antiallodynic effects of ketorolac against BK were not antagonized by quadazocine (1.0 mg/kg), indicating that these effects were not mediated by mu or kappa opioid receptors. Furthermore, relative to morphine, ketorolac displayed a slower onset and a longer duration of action. These findings suggest that the allodynic effects of BK in this procedure were mediated entirely by cyclooxygenase products of arachidonic acid metabolism, such as PGE2.