The paper aimed to compare the efficacy of log odds (LODDS) compared to a classification based on the distribution of involved lymph nodes (pN) and lymph node ratio (LNR).Material was collected retrospectively from an online survey-based database of the Polish Lung Cancer Group and included a group of 17,369 patients who received radical surgical treatment (R0) due to lung cancer.In the whole group the median survival for N0, N1 and N2 was 76.1, 41.7 and 24.2 months, respectively. The median survival for individual LODDS categories (−6,-4], (−4,-3], (−3,-2], (−2,-1], (−1,0], (0,1] and (1,2] was 76.5, 76.3, 71.7, 45.4, 25.0, 19.1 and 17.7 months, respectively. The median survival for LNR in individual categories (0), (0,0.25], (0.25,05], (0.5075] and (0.75,1.0] was 75.6, 40.3, 24.1, 18.8 and 16.4 months, respectively. A multi-variant analysis demonstrated that each LODDS category is an independent prognostic factor: (−4,-3] (HR = 0.982; 95% CI 0.867–1.112; P = 0.775), (−3,-2] (HR = 1.114; 95% CI 0.984–1.262; P = 0.089), (−2,-1] (HR = 1.241; 95% CI 1.080–1.425; P = 0.002), (−1,0] (HR = 1.617; 95% CI 1.385–1.887; P < 0.0001), (0,1] (HR = 1.918; 95% CI 1.579–2.329; P < 0.0001) and (1,2] (HR = 2.016; 95% CI 1.579–2.573; P < 0.0001).Based on LODDS it is possible to discriminate patients with regard to lung cancer stage more effectively compared to pN and LNR classification, and it is also a better classification system.