The results of 300 consecutive thoracic needle biopsies have been evaluated. It is concluded that this is a safe procedure for the investigation of peripheral intrathoracic masses. It should not be performed unless excellent facilities for cytopathology are available. Under these conditions there is a high degree of accuracy in differentiating between malignant and non-malignant lesions, particularly if repeated biopsies are performed when the initial results are unrewarding, equivocal or at variance with the radiographic appearances. In this series there was an imperfect correlation between the cytological and subsequent histological cell type except with squamous carcinomas. The indications are changing with the advent of flexible bronchoscopy and the increasing use of chemotherapeutic regimes. Misleading results may be obtained from biopsy of cavitating lesions and necrotic tumours.