1. The contribution of sensory neurons and mast cells to the oedema evoked by adenosine A1 (N(6)-cyclopentyladenosine, CPA, 3 - 30 nmol site(-1)), A2 (5'N-ethylcarboxamidoadenosine, NECA, 1 - 10 nmol site(-1)) and A3 receptor agonists (N6-[3-iodobenzyl]-N-methyl-5'-carboxiamidoadenosine, IB-MECA, 0.01 - 3 nmol site(-1)) was investigated in the rat skin microvasculature, by the extravascular accumulation of intravenously-injected (i.v.) 125I-albumin. 2. Intradermal (i.d.) injection of adenosine and analogues induced increased microvascular permeability in a dose-dependent manner (IB-MECA > NECA > CPA > adenosine). The non-selective adenosine receptor antagonist theophylline (5 - 50 nmol site(-1)) markedly inhibited adenosine, CPA or NECA but not IB-MECA-induced plasma extravasation. The A1 receptor antagonist 1,3-dipropyl-8-cyclopentylxanthine (DPCPX, 0.3 - 3 micromol kg(-1), i.v.) significantly reduced CPA-induced plasma extravasation whereas responses to adenosine, NECA or IB-MECA were unchanged. The A2 receptor antagonist 3,7-dymethyl-1-proprargylxanthine (DMPX, 0.5 - 50 nmol site(-1)) significantly reduced NECA-induced plasma extravasation without affecting responses to adenosine, CPA and IB-MECA. 3. The tachykinin NK1 receptor antagonist (S)-1-[2-[3-(3,4-dichlorphenyl)-1 (3-isopropoxyphenylacetyl) piperidin-3-yl] ethyl]-4-phenyl-1 azaniabicyclo [2.2.2]octane chloride (SR140333), but not the NK2 receptor antagonist (S)-N-methyl-N[4-acetylamino-4-phenyl piperidino)-2-(3,4-dichlorophenyl)butyl]-benzamide (SR48968), significantly inhibited the plasma extravasation evoked by higher doses of adenosine (100 nmol site(-1)), CPA (100 nmol site(-1)), NECA (1 nmol site(-1)) and IB-MECA (0.1 - 1 nmol site(-1)). In rats treated with capsaicin to destroy sensory neurons, the response to higher doses of adenosine, CPA and NECA, but not IB-MECA, was significantly inhibited. 4. The effects of adenosine and analogues were largely inhibited by histamine and 5-hydroxytryptamine (5-HT) antagonists and by compound 48/80 pretreatment. 5. In conclusion, our results provide evidence that adenosine A1 and A2, but not A3, receptor agonists may function as cutaneous neurogenic pro-inflammatory mediators; acting via microvascular permeability-increasing mechanisms that can, depending on dose of agonist and purine receptor under study, involve the tachykinin NK1 receptor and mast cell amines.