BACKGROUND: Activating FGFR2 mutations are found in 10-16% of primary endometrial cancers and provide an opportunity for targeted therapy. We assessed the safety and activity of dovitinib, a potent tyrosine-kinase inhibitor of fibroblast growth factor receptors, VEGF receptors, PDGFR-β, and c-KIT, as second-line therapy both in patients with FGFR2-mutated (FGFR2(mut)) endometrial cancer and in those with FGFR2-non-mutated (FGFR2(non-mut)) endometrial cancer. METHODS: In this phase 2, non-randomised, two-group, two-stage study, we enrolled adult women who had progressive disease after first-line chemotherapy for advanced or metastatic endometrial cancer from 46 clinical sites in seven countries. We grouped women according to FGFR2 mutation status and gave all women dovitinib (500 mg per day, orally, on a 5-days-on and 2-days-off schedule) until disease progression, unacceptable toxicity, death, or study discontinuation for any other reason. The primary endpoint was proportion of patients in each group who were progression-free at 18 weeks. For each group, the second stage of the trial (enrolment of 20 additional patients) could proceed if at least eight of the first 20 treated patients were progression free at 18 weeks. Activity was assessed in all enrolled patients and safety was assessed in all patients who received at least one dose of dovitinib. The completed study is registered with ClinicalTrials.gov, number NCT01379534. FINDINGS: Of 248 patients with FGFR2 prescreening results, 27 (11%) had FGFR2(mut) endometrial cancer. Between Feb 17, 2012, and Dec 13, 2013, we enrolled 22 patients in the FGFR2(mut) group and 31 patients in the FGFR2(non-mut) group. Seven (31·8%, 95% CI 13·9-54·9) patients in the FGFR2(mut) group and nine (29·0%, 14·2-48·0) in the FGFR2(non-mut) group were progression-free at 18 weeks. On the basis of predefined criteria, neither group continued to stage two: seven (35%) of the first 20 patients in the FGFR2(mut) group were progression free at 18 weeks, as were five (25%) of the first 20 in the FGFR2(mut) population. Rates of treatment-emergent adverse events were similar between groups and events were most frequently gastrointestinal. Overall, the most common grade 3 or 4 adverse events suspected to be related to the study drug were hypertension (nine patients; 17%) and diarrhoea (five; 9%). The most frequently reported serious adverse events suspected to be related to study drug were pulmonary embolism (four patients; 8%), vomiting (four; 8%), dehydration (three; 6%), and diarrhoea (three; 6%). Only one death was deemed to be treatment-related: one patient in the FGFR2(non-mut) group died from cardiac arrest with contributing reason of pulmonary embolism (grade 4, suspected to be study drug related) 4 days previously. INTERPRETATION: Second-line dovitinib in FGFR2(mut) and FGFR2(non-mut) advanced or metastatic endometrial cancer had single-agent activity, although it did not reach the prespecified study criteria. Observed treatment effects seemed independent of FGFR2 mutation status. These data should be considered exploratory and additional studies are needed. FUNDING: Novartis Pharmaceuticals.