Objectives: Selection criteria for radical surgery in malignant pleural mesothelioma (MPM) and related clinical trials remain controversial. The relative importance of nodal metastases and the need for pre-operative nodal staging are undetermined., Methods: From a prospective database, we identified 212 patients with non-sarcomatoid MPM (160 epithelioid and 52 biphasic). A total of 127 patients underwent extrapleural pneumonectoctomy (EPP) and 85 lung-sparing total pleurectomy (LSTP) with lymphadenectomy. We investigated the effect of nodal burden and distribution in survival by testing for differences between N0, N1 and N2 disease and constructing a theoretical model dividing the patients into four groups according to diseased nodes identified in the surgical specimen: Group 0, no nodal disease; Group CM, nodes accessible by cervical mediastinoscopy (CM): Stations 2, 3a, 4 and 7; Group EBUS/EUS, nodes accessible by endobronchial (EBUS) or endoscopic (EUS) ultrasound: Stations 2, 3a, 4 and 7-11. Group EM, extramediastinal nodes not accessible by CM or EBUS/EUS: Stations 5, 6, internal mammary, pericardial and diaphragmatic lymph nodes., Results: There was no difference in overall median survival between EPP and LSTP [15.6, SE 1.8, 95% confidence interval (CI) 12-19 months versus 13.4, SE 2.3, 95% CI 9-18 months, P=0.41]. Patients with N0 disease (n=94) had the best prognosis: median survival was 19.6 months (SE 3, 95% CI 13.2-26) versus 12 months for the 19 patients with N1 (SE 1.5, 95% CI 9-15) and 13.6 months for 99 patients with N2 (SE 1.7, 95% CI 10-17), P=0.015. Subgroup analysis of patients with nodal metastases revealed no significant survival difference between group CM and group EBUS/EUS: achieving maximum theoretical diagnostic yield CM could detect 63 (54%) of patients with nodal disease and the median survival of this group was 13.6 months (SE 2, 95% CI 9.6-17.6). EBUS/EUS could detect an additional 30 cases (26%) with survival of 11.3 months (SE 1, 95% CI 9-13.6). The survival in group EM (25 cases, 21%, median survival 18.7 months, SE 6, 95% CI 7-30) was significantly better than groups CM or EBUS/EUS, P=0.002., Conclusions: There is a strong case for routine CM as a method of prognostic staging in all patients undergoing radical surgery for MPM. The addition of EUS staging and the detection of nodal metastases inaccessible to mediastinoscopy had no prognostic benefit.