Patients with recurrent/metastatic uterine leiomyosarcoma (U-LMS) have a dismal prognosis. This phase II study aims to evaluate trabectedin efficacy and safety in advanced U-LMS. Eligible patients had received ≥ one line of chemotherapy. Gemcitabine ± docetaxel naive patients were randomised to Arm A: trabectedin 1.3 mg/m or calibration Arm B: gemcitabine 900 mg/m and docetaxel 75 mg/m . Patients who had already received gemcitabine ± docetaxel directly entered Arm A. Primary end-point: 6-month progression-free rate (PFS-6). The null hypothesis that the true PFS-6 = 14% was tested against a one-sided alternative. This design yielded a 5% type I error rate and 90% power when the true PFS-6 is 25%. Overall, 126 patients entered Arm A (45 from randomisation and 81 directly) and 42 Arm B. Arm A patients characteristics: median age = 57; ≥2 previous chemotherapy lines = 37.4%; metastatic disease = 93%. The study met the condition for trabectedin activity: PFS-6 = 35.2% (95% CI: 26.2-45). No difference in PFS by the number of previous chemotherapy lines emerged. Median OS = 20.6 months (IQR: 8-36.4). In Arm B, the PFS-6 = 51.5% (95% CI: 33.5-69.2). No toxic deaths occurred. In Arm A, only 4 patients interrupted treatment for toxicity. Trabectedin is active and well tolerated, retaining similar efficacy across one to three previous lines of chemotherapy.