1. Nitric oxide has been described either as pronociceptive or antinociceptive. In this investigation, using an electronic pressure-metre, the intradermal and the subcutaneous effects of prostaglandin E(2) (PGE(2)) and agents that mimic or inhibit the arginine/NO/cGMP pathway were compared. 2. The hypernociceptive effect of the intradermal injection of PGE(2) (100 ng) was immediate, peaking within 15-30 min and returning to basal values in 45-60 min. The subcutaneous injection of PGE(2) induced a hypernociception with a delayed peak (3 h) plateauing for 4-6 h. 3. Intradermal administration of 3-morpholino-sydnonimine-hydrochloride (SIN-1) enhanced, while its subcutaneous administration inhibited, subcutaneous hypernociception induced by PGE(2). This inhibition was prevented by ODQ (8 micro g) but not by NG-monomethyl-L-arginine (L-NMMA) (50 micro g). 4. Intradermal but not subcutaneous administration of L-arginine (1-100 micro g), SIN-1 (1-100 micro g) and dibutyrylguanosine 3':5'-cyclic monophosphate (db cGMP) (0.1-100 micro g) induced an early (15-30 min) dose-dependent hypernociceptive effect. Intradermal pretreatment with NG-monomethyl-L-arginine (L-NMMA; 50 micro g) inhibited the hypernociception induced by L-Arg (10 micro g), but not that induced by SIN-1 (10 micro g) or db cGMP (10 micro g). 5. Intradermal injection of ODQ (8 micro g) antagonized the hypernociception induced by L-arginine and SIN-1, but not that induced by db cGMP. 6. Considering (a) the different time course of intradermal and subcutaneous PGE(2)-induced hypernociception, (b) the opposite nociceptive effect of intradermal and subcutaneous administration of SIN-1 (db cGMP) as well as the arginine/NO/cGMP pathway, the existence of different subsets of nociceptive primary sensory neurons in which the arginine/NO/cGMP pathway plays opposing roles is suggested. This hypothesis would explain the apparent contradictory observations described in the literature.