In the present study we attempted to find out if the rostroventromedial medulla (RVM) or the periaqueductal gray (PAG) might contribute to chronic allodynia induced by unilateral ligation of two spinal nerves in the rat. Lidocaine was microinjected in the RVM or PAG and allodynia was quantitatively determined by measuring the hindlimb withdrawal thresholds to mechanical stimulation of the paw. For comparison, lidocaine was also injected systemically (s.c.). Lidocaine in the RVM produced a dose-related (20 and 40 micrograms) antiallodynic effect. Lidocaine (20 micrograms) in the PAG produced identical antiallodynic effect as in the RVM. With systemic administrations of lidocaine, a considerably higher dose (> > 40 micrograms) was needed to produce a significant antiallodynic effect. Naloxone, an opioid-antagonist (1 mg/kg s.c.), did not attenuate the antiallodynic effect of lidocaine in the RVM. An antiallodynic dose of lidocaine (20 micrograms) in the RVM or the PAG did not influence the withdrawal response in the unoperated hindlimb nor the heat-induced tail-flick reflex. The results indicate that the RVM and the PAG have a facilitatory influence on the spinal segmental mechanisms underlying chronic allodynia. The selective attenuation of allodynia induced by lidocaine in the RVM and the PAG is independent of opiate receptors, and it can not be explained by a systemic spread of the drug.