Abnormal uterine bleeding is the major reason for discontinuing long-term progesterone-only contraceptives (LTPOCs). Prior studies demonstrated that endometria exposed to the LTPOC, Norplant, display aberrant angiogenesis, leukocyte infiltration, and hypoxia-associated impaired blood flow. Paradoxically, human endometrial stromal cells (HESCs) of these specimens exhibit elevated expression of tissue factor (TF), the primary initiator of hemostasis via thrombin generation. The current study demonstrates that TF levels are also elevated in HESCs that are decidualized after insertion of Mirena, an intrauterine system that releases levonorgestrel directly into the endometrial canal and produces elevated perivascular levels of the proinflammatory and angiognenic cytokine IL-8. Because bleeding, inflammation, and ischemia-associated increased vascular permeability enhance access of plasma factor VII to HESC-expressed TF to generate thrombin, we evaluated the effects of steroids, thrombin, and hypoxia on HESC expression of IL-8. Confluent HESCs were incubated in a serum-containing medium for 7 d with vehicle control or estradiol (E(2)) plus medroxyprogesterone acetate (MPA). The medium was then exchanged for corresponding defined medium with and without thrombin, and the cultures were incubated in parallel for up to 48 h in a standard incubator (normoxia) or a sealed chamber at 0-1% O(2) (hypoxia). Under normoxia, immunoreactive IL-8 levels in the conditioned medium were reduced to one-third of control levels during decidualization with E(2)+MPA (P < 0.05; n = 5). In E(2)+MPA-treated cultures, thrombin (0.1 U/ml to 2.5 U/m) elicited a dose-dependent reversal of this inhibition, elevating IL-8 up to 60-fold (P < 0.05; n = 5) for more than 24 h and steady-state IL-8 mRNA levels by 3-fold for 3 h. The specific inactivator, hirudin, blocked most of the effects of thrombin, whereas TRAP-14, an agonist of the protease-activated receptor for thrombin, enhanced IL-8 output. In the absence of thrombin, hypoxia elevated IL-8 output 5-fold in E(2)+MPA-treated HESCs (P < 0.02, n = 4), with thrombin exerting additive effects. In contrast to its effects in progestin-treated HESCs, hypoxia did not elevate IL-8 output in control cultures. This study suggests that inhibition of IL-8 expression in decidualized HESCs contributes to the antiinflammatory milieu of the luteal phase. However, LTPOC-induced hypoxia and excess thrombin generation enhance IL-8 expression in decidualized HESCs, thereby eliciting aberrant angiogenesis and inflammation that promote the onset of abnormal uterine bleeding.