Purpose: To evaluate the oncological impact of postponing radical cystectomy (RC) to allow further conservative therapies prior to progression in a large multicentre retrospective cohort of T1-HG/G3 patients initially treated with BCG. Methods: According to the time of RC, the population was divided into 3 groups: patients who did not progress to muscle-invasive disease, patients who progressed before radical cystectomy and patients who experienced progression at the time of radical cystectomy. Clinical and pathological outcomes were compared across the three groups. Results: Of 2451 patients, 509 (20.8%) underwent RC. Patients with tumors > 3 cm or with CIS had earlier cystectomies (HR = 1.79, p = 0.001 and HR = 1.53, p = 0.02, respectively). Patients with tumors > 3 cm, multiple tumors or CIS had earlier T3/T4 or N + cystectomies. In patients who progressed, the timing of cystectomy did not affect the risk of T3/T4 or N + disease at RC. Patients with T3/T4 or N + disease at RC had a shorter disease-specific survival (HR = 4.38, p <0.001), as did patients with CIS at cystectomy (HR = 2.39, p <0.001). Patients who progressed prior to cystectomy had a shorter disease-specific survival than patients for whom progression was only detected at cystectomy (HR = 0.58, p = 0.024) Conclusions: Patients treated with RC before experiencing progression to muscle-invasive disease harbor better oncological and survival outcomes compared to those who progressed before RC and to those upstaged at surgery. Tumor size and concomitant CIS at diagnosis are the main predictors of surgical treatment while tumor size, CIS and tumor multiplicity are associated with extravesical disease at surgery. © 2018, Springer-Verlag GmbH Germany, part of Springer Nature.