From 1982 to 1993, four hundred and sixty transthoracic biopsies were performed to diagnose lung lesions. 258/460 examinations (56%) were made under fluoroscopic and 202/460 exams (44%) under CT guidance. All the instrumental examinations were made with fine cyto/histologic needles (21-22G); the pathologist was always present. Lesion features and biopsy modalities were analyzed statistically to assess which factors correlate with biopsy yield and complications. Sampling efficacy averaged 85%, specific cell typing was 96% and diagnostic accuracy was 97%. The radiologic features of the lesions (masses or inhomogeneous opacities) and their size, needle type (cyto/histologic needles), the number of punctures per biopsy and examination time were statistically influential factors on biopsy accuracy. Complications were observed in 31% of cases, 5% of them consisting of pneumothorax cases requiring surgical drainage. The factors bearing a statistical influence on complications were lesion site and size, lung parenchymal alterations, needle course depth, scissure plane crossing, the patient's position during biopsy and examination time. Finally, CT allowed smaller and deeper lesions to be studied than fluoroscopy, with the same diagnostic yield.