BACKGROUND: Identification of the histologic cell type of lung cancer is important because it is related to the treatment modality and prognosis. Currently, many diagnostic methods are used to determine the cell type. We have studied to evaluate the histologic accuracy of each diagnostic methods in lung cancer. METHOD: 168 cases were analysed retrospectively, who underwent curative thoracotomy for lung cancer in Seoul National University Hospital from January, 1994 to Feburary, 1997. Sputum cytology, percutaneous needle aspiration(PCNA), percutaneous needle biopsy(PCNB), bronchial washing cytology and flexible bronchoscopic biopsy were evaluated respectively. The kappa coefficient was applied to evaluate the degree of concordance between the histologic diagnosis obtained by each methods and the one derived from thoracotomy. RESULTS: 1. The kappa value in sputum cytology was 0.86 In the 34 cases with squamous cell carcinoma(SQ), 32 cases were finally diagnosed as having such by thouacotomy. 7 of the 8 cases with adenocarcinoma(AD) turned out to have the same. 2. The kappa value in PCNA was 0.51. In the 31 cases with SQ, only 14 cases were finally diagnosed as having such. All of the 3 cases with small cell carcinoma(SC) turned out to have the same. 3. The kappa value in PCNB was 0.77. The diagnosis was correct in 13 of the 16 patients with SQ and in 30 of the 32 cases with AD. 4. The kappa value in bronchial washing cytology was 1.0. In all of the 29 cases with SQ and all of the 7 cases with AD, the diagnosis was correct. 5. The kappa value in flexible bronchoscopic biopsy was 0.77. The diagnosis was correct in 51 of the 52 cases with SQ and in 1 of the 2 cases with SC. CONCLUSION: The concordance rate with the final histologic diagnosis in sputum cytology, PCNB, bronchial washing cytology and flexible bronchoscopic biopsy were excellent (kappa >or=0.75), while that in PCNA was fair (kappa=0.53). Because PCNA showed lower concordance rate than other diagnostic methods, PCNA is recommended to perform with PCNB.