Objective: To assess the effect of lymph node dissection (LND), number of removed nodes (NRN), and number of positive nodes (NPN), on cancer-specific mortality (CSM) in contemporary vs historical patients with pT 2-3 N any M 0 renal cell carcinoma (RCC) treated with radical nephrectomy (RN). Patients and Methods: Within the Surveillance, Epidemiology, and End Results database (2001-2013), we identified patients with non-metastatic pT 2-3 N any RCC who underwent RN with or without LND. Kaplan-Meier analyses and multivariable Cox regression models with propensity score weighting for inverse probability of treatment were used. Results: Of 25 357 patients, 24.8% underwent LND (2001-2007: 3 167 patients vs 2008-2013: 3 133 patients). The median NRN was 3 (interquartile range [IQR]: 1-7). Positive nodes were identified in 17.1%: 9.3% of pT 2 and 21.6% of pT 3 patients, who underwent LND. The median NPN was 2 (IQR: 1-3). In multivariable models, LND did not decrease CSM (hazard ratio [HR] 1.29; P < 0.001). LND extent, defined as NRN, did not decrease CSM (HR 0.94; P = 0.3). Finally, multivariable models testing the effect of NPN showed increased CSM in pT 3 but not in pT 2 patients (HR 1.29 and 1.58, P = 0.02 and P = 0.1, respectively). NRN exerted a protective effect on CSM in patients with positive nodes (HR 0.98; P = 0.007). Conclusion: In contemporary and historical patients LND or its extent do not protect from CSM. However, the NPN increases the rate of CSM in pT 3 patients. Consequently, LND and its extent appear to have little if any therapeutic value in pT 2-3 N any M 0 patients, besides its prognostic impact. High-risk non-metastatic patients may represent a target population for a multi-institutional prospective trial. © 2017 BJU International.