The histologic types of lung cancer cases diagnosed in 1979-1980 (n=799) and 1987 (n=587) were independently reviewed by two pathologists in order to investigate the reproducibility of the diagnosis of the histologic type when the WHO classification (1981) was used. The specimens from 354 surgical cases and biopsy or cytology specimens from 1032 non-surgical cases were reviewed. The inter-observer agreement was 87.9% (kappa=0.79) for surgical cases and 81.4% (kappa=0.72) for non-surgical cases. When compared to the original diagnosis, the agreement was 86.8% (kappa=0.78) for surgical and 86.4% (kappa=0.79) for non-surgical cases in 1979-1980 and the agreement was 92.8% (kappa=0.87) for surgical and 89.1% (kappa=0.83) for non-surgical cases in 1987. By histologic type, no difference in the agreement was observed except for large cell carcinoma. The distribution of histologic types after the review differed only slightly (less than 6%) from the original distribution. This suggests that in Osaka, Japan, the diagnosis based on the WHO classification (1981) had only a limited influence on the distribution of histologic types, and is not a major reason for the changing trends in lung cancer incidence by histologic type.