2-[18F] Fluoro-D-deoxyglucose positron emission tomography (FDG-PET) is an appropriate diagnostic procedure for staging lung cancer. However, accurate evaluation of lymph node (LN) metastases by PET is controversial owing to false-positive/-negative FDG uptake results. The prognostic significance of both false-negative and false-positive LNs on FDG-PET remains to be determined. A total of 235 patients with lung cancer were retrospectively analyzed. Maximum standardized uptake values (SUVmax) of the lymph nodes were compared with pathological LN metastases to correlate PET findings with clinicopathological variables and patients' outcomes. When SUVmax ≥ 4 was defined as PET-positive for LN metastasis, sensitivity, specificity, and accuracy were 46.0%, 79.5%, and 72.3%, respectively. False-negative cases and pathological n0 cases were significantly younger, had primary tumors that were smaller or lower SUVmax, and adenocarcinomas compared with false-positive and pathological n+ cases. The difference in survival time between patients with abnormal FDG uptake in the LN and those without was larger than that between pathological LN metastases and no pathological metastases in patients with adenocarcinoma. Multivariate analysis by the Cox proportional hazard model identified smoker, EGFR/ALK negative and LN positive on PET as significant adverse prognostic factors, rather than pathological n-stage. Abnormal FDG uptake in the LN is an important prognostic factor. Increased glucose metabolism on FDG-PET appears to be a more efficient postoperative prognostic marker than pathological n-stage in patients with lung cancer.