Fine needle aspiration biopsy (FNAB) and punch biopsy (PB) are reliable methods of establishing a morphologic diagnosis in thoracic lesions. However, some reservations exist concerning the diagnostic accuracy of and indications for both methods. Therefore, we evaluated the sensitivity, specificity, typing accuracy and complication rates of both methods. We present a six-year experience with 501 thoracic FNABs and PBs in 482 patients. To examine site-specific differences, we evaluated three different compartments: lung, mediastinum and hilum. In 457 cases the final outcome was known for evaluating the accuracy of the cytologic or histologic diagnoses. FNAB was used most often in lung (81.8%) and hilar lesions (87.3%), whereas PB was used mostly in mediastinal (67.9%) and pleural lesions or if a mesenchymal lesion was suggested radiologically but never in foci below 20 mm in diameter. Our complication rate was 21.3% for FNAB and 4.6% for PB. The most frequent complications were pneumothorax, one hematothorax and intercostal neuralgia. The overall sensitivities of the biopsy methods were equal (FNAB, 98.4%; PB, 98%), but the typing accuracy was better for PB than FNAB (87.2% vs. 83.5%). In the hilum the sensitivities of FNAB and PB were 94.6% and 85.7%, respectively, and for the lungs, 99% and 98.2%. In the mediastinum the sensitivity was 100% for both methods. There were false-positive diagnoses in 5% with FNAB of the lung due to misinterpretation of regenerating epithelium and hamartochondroma and a 0.1% rate of false-negative diagnoses as a result of misplacement of the cannula, leading to inflammation, infarction or scarring. Our data indicate that FNAB is the method of choice in pulmonary and hilar lesions because of the similar diagnostic accuracy. Mediastinal and pleural lesions and presumed mesenchymal tumors should be sampled with PB because the typing accuracy of FNAB is insufficient in these cases.