The stagnant mortality rates for metastatic urothelial cancer (UC) have provoked efforts to find novel treatments. To test the utility of the extirpative surgery for primary tumor as an option for these patients, we investigated the perioperative and oncologic outcomes of surgery for primary tumors in metastatic UC patients.We reviewed the medical records of 130 metastatic UC patients (bladder: 88, upper tract UC: 42) at diagnosis from November 2005 to November 2016. A total of 56 patients (surgery group) underwent chemotherapy with extirpative surgery for the primary tumor, and 74 patients (non-surgery group) received chemotherapy. We evaluated perioperative outcomes, cancer-specific survival (CSS), and overall survival (OS) using Kaplan-Meier methods and factors related to OS and CSS using Cox regression models.Surgery group showed similar perioperative outcome and postoperative complications to those previously reported in UC patients without metastasis, and fewer urinary complications than non-surgery group. Surgery group showed better oncological outcomes than non-surgery group for median CSS (16.0 vs 10.0 months, P = 0.014) and median OS (14.0 vs 9.0 months, P = 0.043). Multivariate analysis showed Eastern Cooperative Oncology Group performance status and metastasis to liver as significant predictors of CSS and OS. Surgery was not related with OS, but a significant predictor of CSS.Extirpative surgery for primary tumor in metastatic UC can be feasible and it might have survival benefits, especially those patients with a tolerable general condition and no liver metastasis. In addition, LT reduces the possibility of a surgical procedure towing to urinary complications.