1. Ligustrazine (tetramethylpyrazine, TMP) is a vasodilator that has been reported to have pulmonary selective properties in vivo, but not in vitro. Although TMP is generally described as being endothelium-independent, we provide evidence here that TMP may have an endothelium-dependent and nitric oxide (NO)-mediated mechanism in pulmonary arteries that could predominate at concentrations used therapeutically in China. 2. The study was performed on isolated pulmonary (1-2 mm i.d.), intrapulmonary (200-850 microns) and mesenteric (200-400 microns) arteries of the rat using a Mulvaney-Halpen small vessel myograph, following preconstriction with phenylephrine (PE, 10 microM), prostaglandin F2 alpha (PGF2 alpha, 100 microM), or 75 mM K+ (KPSS, equimolar substitution for Na+). Values are shown as mean +/- s.e.mean, or for EC50S as mean [+/-95% confidence limits]. 3. TMP caused a concentration-dependent relaxation against all three agonists in both large (1.56 +/- 0.04 mm) and small (399 +/- 20 microM) pulmonary arteries; it was more potent in small compared to large arteries constricted with PE or PGF2 alpha (P < 0.05), but not those constricted with KPSS. The NO synthase (NOS) inhibitor, NG-monomethyl-L-arginine (L-NMMA, 100 microM) caused a significant shift to the right of these relationships, such that the EC50 for TMP in large pulmonary arteries constricted with PE increased from 522 [+130, -104] microM (n = 12) to 1828 [+395, -325] microM (n = 6, P < 0.01). Both removal of the endothelium and methylene blue (10 microM) had similar effects. 4. L-Arginine substantially reduced the EC50 for TMP in pulmonary arteries; in the presence of 400 microM L-arginine the EC50 for TMP in large arteries constricted with PE was 14.7 [+21.0, -8.6] microM, (n = 6, P < 0.001), and with 10 microM L-arginine 96.7 [+45.1, -30.7] microM, (n = 6, P < 0.001). Similar effects were seen in small arteries. L-Arginine had no effect in the absence of an endothelium. D-Arginine was ineffective, and inhibition of L-arginine uptake with L-lysine blocked the action of L-arginine. L-Arginine (400 microM) had no significant effect on TMP-induced relaxation in mesenteric arteries (n = 5). 5. L-Arginine itself caused a concentration-dependent relaxation in intrapulmonary arteries (639 +/- 34 microM) constricted with PE, reaching a maximum relaxation around 100-400 microM (42.4 +/- 3.0%, n = 16), but this was independent of the endothelium. TMP (10 and 100 microM) significantly enhanced the relaxation to L-arginine, with a maximum relaxation in the presence of 100 microM TMP of 81.7 +/- 6.2% (n = 5, P < 0.01), but the effect of TMP was entirely dependent on the endothelium. A similar effect was observed in PGF2 alpha-constricted pulmonary arteries. 6. These results show that TMP stimulates NO production at low concentrations in pulmonary arteries, via an apparently novel endothelium-resident mechanism that is dependent on exogenous L-arginine. Normal plasma L-arginine levels of around 150 microM would allow this mechanism to be maximally activated. As mesenteric arteries do not seem to express the mechanism to any significant extent, at low concentrations TMP would be effectively selective to the pulmonary vasculature, and may thus have potential as a therapeutic agent in pulmonary vascular disease.