We prospectively studied results of 103 consecutive transthoracic needle biopsies of lung lesions suspicious for malignancy to determine if lesion size, depth within the lung, hyperinflation, second needle passes, or 100 percent oxygen breathing influenced the incidence of pneumothorax. Thirty-eight patients (37 percent) developed pneumothorax. Ten (10 percent) required tube thoracostomy for re-expansion. Five of the ten requiring the chest tube had clinically severe obstructive lung disease. In all patients, greater depth (D) in centimeters of needle penetration significantly increased the probability (p) of pneumothorax (p much less than .001) and can be estimated by the equation: (formula see text) The estimated probability of pneumothorax is 13 percent at 1 cm, 49 percent at 4 cm, and 86 percent at 7 cm. An increase in total lung capacity above predicted added additional risk of pneumothorax (p less than .02). Oxygen breathing did not significantly reduce the incidence of pneumothorax, but may reduce size by increasing the rate of reabsorption. The other factors had little influence. We conclude that the more central location of the lesion and pre-existing lung hyperinflation determine the risk of pneumothorax.