OBJECTIVES: Overlapping survival curves for N1b (multiple N1 stations), N2a2 (single N2 station + N1 involvement) and N2a1 (skip N2 metastasis) limit the current tumour-node-metastasis (TNM) node (N) subclassification for node involvement. We validated externally the proposed subclassification. METHODS: Clinical records from a multicentric database comprising 1036 patients with pulmonary adenocarcinoma (ADC) or squamous cell carcinoma with N1/N2 involvement who underwent, from January 2002 to December 2014, complete lung resections were retrospectively reviewed. Patients were categorized according to the 8th TNM N subclassification proposal. Histological type, number of resected nodes (RN) and adjuvant therapy (ADJ) were considered limiting factors. RESULTS: No difference in the 5-year overall survival (-OS) was noted between N1b and N2a1 (49.64.8 P = 0.72); instead, the 5-year-OS was significantly improved in patients with squamous cell carcinoma (63b vs 30.7a1, P = 0.04). In patients with ADC, the 5-year-OS was better in those with N2a1 than with N1b (50.67.5 P = 0.09). When we compared N1b with N2a2, the 5-year-OS was statistically significant (49.62.8 P = 0.02); considering only patients with squamous cell carcinoma (635.8 P = 0.003), RN textgreater10 (63.25.3 P = 0.05) and without ADJ (56.44.5 P = 0.02), the 5-year-OS was significantly different. Differences were not significant for ADC, RN textless10 and ADJ. Finally, the 5-year-OS was statistically significant when we compared N2a1 with N2a2 of the total cohort (44.82.8 P = 0.04), in ADC (5-year-OS 50.66.5 P = 0.04) and RN textgreater10 (5-year-OS 49.82.1 P = 0.03) without ADJ. CONCLUSIONS: Histological type, ADJ and RN are relevant prognostic factors in N + non-small-cell lung cancer. Considering these results, we may better interpret the prognosis prediction limits of the proposed 8th TNM subclassification for the N descriptor.