Abstract Background The prognosis for patients with lymph node (LN)–positive bladder cancer (BCa) is likely affected by the extent of lymphadenectomy in radical cystectomy (RC) cases. Specifically, the prognostic significance of the LN density (ratio of positive LNs to the total number removed) has been demonstrated. Objective To evaluate the prognostic signature of lymphadenectomy variables, including the LN density, for a large, multicentre cohort of RC patients with LN-positive BCa. Design, setting, and participants The clinical and histopathologic data from 477 patients with LN-positive urothelial BCa (pN1–2) were analysed. The median follow-up period for all living patients was 28 mo. Measurements Multivariable Cox regression analysis was used to test the effect of various pelvic lymph node dissection (PLND) variables on cancer-specific survival (CSS) based on colinearity in various models. Results and limitations The median number of LNs removed was 12 (range: 1–66), and the median number of positive LNs was 2 (range: 1–25). Two hundred ninety (60.8%) of the patients presented with stage pN2 disease. The median and mean LN density was 17.6% and 29% (range: 2.3–100), respectively, where 268 (56.2%) and 209 (43.8%) patients exhibited am LN density of ≤20% and >20%, respectively. In separate multivariable Cox regression models adjusted for age, sex, pTN stage, grade, associated Tis, and adjuvant chemotherapy, the interval-scaled LN density (hazard ratio [HR]: 1.01; p = 0.002) and the LN density, ordinal-scaled by 20% (HR: 1.65; p < 0.001) exhibit independent effects on CSS. In addition, an independent contribution appears from the pT but not the pN stage. Limitations include surgeon selection bias when determining the extent of lymphadenectomy. Conclusions Our results support the prognostic relevance of LN density in patients with LN-positive BCa, where a threshold value of 20% stratifies the population into two prognostically distinct groups. Before LN density is integrated into the clinical decision-making process, these results should be validated by prospective studies with defined LN templates and standardised histopathologic methods.