The predictive value of lymph node micrometastasis, detected by immunohistochemical or genetic methods, is well appreciated in terms of prognosis. However, a major problem is high false-positive rates, because most methods focus on cytokeratin, which is a component not only of carcinoma but also normal epithelial and nonepithelial cells. Mutant allele-specific amplification (MASA) can detect DNAs derived from cancer cells itself, reportedly with high sensitivity. It was, therefore, used with nested-PCR using p53 or K-ras mutation for analysis of lymph node micrometastasis in non-small cell lung carcinoma (NSCLC) patients in the present study, in comparison with the immunohistochemical method using an anti-cytokeratin reagent for the same samples. Lymph nodes from 31 NSCLC patients with p53 and K-ras mutated tumors (30 and 1, respectively) staged as pathological (p)-T1-4 N0-1 and M0 were examined. Genetic and immunohistochemical methods demonstrated positive reactions in 34 (15%) and 61 (27%) of 229 lymph nodes, respectively (9 cases, 29%, and 24 cases, 77%). The concordance with the two methods was 77%, but 13 (39%) of 34 genetically positive lymph nodes could not be detected by immunohistochemistry (IHC). Of 22 cases with p-N0 disease, 6 (27%) were genetically positive in hilar and/or mediastinal lymph nodes, and 4 (67%) of them died after cancer relapse. In contrast, none of the patients without micrometastasis died of cancer (P < 0.001, log rank analysis). Of the same p-N0 patients, 17 (77%) were positive by IHC, and 4 (24%) of them died of cancer, whereas 5 negative patients did not suffer cancer relapse. Survival did not significantly differ between cases positive and negative (P = 0.246) by IHC. According to the g-N (N factor restaged by a genetic method), patients with g-N1 and g-N2 disease had a shorter survival than those with g-N0 disease (P = 0.042 and P < 0.001, respectively). However, no significant difference was observed with grading by IHC. Thus, detection of micrometastasis in regional lymph nodes with the MASA method, in other words with a carcinoma-specific marker, is of greater prognostic significance for early stage NSCLC patients than immunohistochemical results. This approach should facilitate selection of patients for whom postoperative adjuvant chemotherapy should be performed.