To compare the accuracy in detecting variant histologies (VH) at transurethral resection of bladder (TURB) and radical cystectomy (RC) specimen among tertiary referral centres, in order to investigate potential reasons of discrepancies from the pathological point of view. Clinical and histopathological data of TURB specimen and subsequent cystectomy specimen of 3,445 RC candidate patients have been retrospectively collected from 24 tertiary referral centres between 1980 and 2021. VH considered in the analysis were pure squamous cell carcinoma, urothelial carcinoma with squamous differentiation, pure adenocarcinoma, urothelial carcinoma with glandular differentiation, micropapillary bladder cancer (BCa), neuroendocrine BCa, and other variants. The degree of agreement between TURB and RC concerning the identification of VH was expressed as concordance, classified according to Cohen's kappa coefficient. A VH was reported in 17% of TURB specimens, 45% of which were not confirmed in RC. The lowest concordance rate was reported for micropapillary BCa with 11 out of 18 (61%) centres reporting no agreement, whereas neuroendocrine BCa achieved the highest concordance rate with only 3 centres (17%) reporting no agreement. Our results shows that even among centres with the advantage of a referent uropathologist the micropapillary variant is characterized by scarce accuracy between TURB and RC. Differences in TURB specimen acquisition by the urologist and in sampling methods among different centres are the main limitations of the study. Accuracy of TURB in detecting VH is poor for certain VH, in particular for micropapillary BCa, with evident variation among centres. Novel diagnostic tools are required to better identify these VH and drive patients toward a personalized treatment.