To evaluate the growing tendency in recent years to attribute more diagnostic reliability to cytologic methods, we investigated the accuracy of cytologic typing in specimens obtained from bronchopulmonary material by five different clinical sampling methods, comparing the cytologic diagnoses with the known histologic diagnoses. The study consisted of 232 cytologic specimens from 157 cases of primary lung cancer. Of the 232 specimens, 173 (75%) were correctly typed and 59 (25%) incorrectly typed with respect to the appropriate histologic diagnoses. When all sampling methods were considered together, the study demonstrated that well-differentiated epidermoid carcinoma and "oat cell" and spindle-polygonal anaplastic carcinomas yielded high cytologic typing accuracies. In poorly differentiated tumors, bronchioloalveolar cell carcinoma and bronchogenic adenocarcinoma, the correct cytologic typing was much lower. The different tumor types and their degrees of differentiation seem to be the decisive factors in cytologic typing accuracy. The findings of this study were compared with those of others and were found to be consistent with the results of even larger series of cases. For some types the typing accuracy was higher than that reported in other series, whereas for other types, e.g., adenocarcinomas, it was lower.