Background: Obstructive sleep apnea (OSA) is a prevalent disorder causing hypertension. Endothelial dysfunction appears to underlie development of hypertension. It is not known whether hypoxia during sleep is necessarily the prerequisite process for endothelial dysfunction and hypertension in OSA. We therefore examined the relationship between endothelial-dependent vasodilatory capacity, hypoxia and circulating angiogenesis inhibitors in OSA. Methods and results: We studies 95 subjects with and without OSA and hypertension. Endothelialdependent vasodilation was assessed using brachial artery flow-mediated vasodilation method (FMD). Plasma angiogenesis inhibitors, endoglin (sEng) and fms-like tyrosine kinase-1 (sFlt-1), were measured using ELISA. The apneaehypopnea indexes were 41 5 and 48 4 events/hr in normotensive OSA (NOSA) and hypertensive OSA (H-OSA), respectively, indicating severe OSA. The sleep time spent with SaO2 < 90% (T < 90%) were 34 8 and 40 9 min, respectively. FMD was markedly impaired in H-OSA (8.0% 0.5) compared to N-OSA (13.5% 0.5, P < 0.0001), H-non-OSA (10.5% 0.8, P < 0.01), and N-non- OSA (16.1% 1.0, P < 0.0001). There was no correlation between T < 90% and FMD. Both OSA groups had elevated levels of sFlt-1 (62.4 5.9 and 63.9 4.7 pg/ml) compared to N-non-OSA (32.1 6.5, P ¼ 0.0008 and P ¼ 0.0004, respectively) and H-non-OSA (41.2 7.0, P < 0.05 and P ¼ 0.03, respectively). In contrast, sEng was only elevated in H-OSA (4.20 0.17 ng/ml) compared with N-OSA (3.64 0.14, P ¼ 0.01) and N-non-OSA (3.48 0.20, P ¼ 0.01). There was a modest but statistically significant inverse correlation between sEng and FMD in only H-OSA group (r ¼ 0.38, P < 0.05). Conclusion: These data show that patients with OSA and hypertension have marked impairment of FMD, independent of hypoxia exposure, which is associated with increased sEng.