The decision to insert an intercostal drain in chest injury must occasionally be made without a chest radiograph. A prospective analysis of the reliability of physical examination in penetrating pleural injuries was undertaken. A total of 51 consecutive patients were examined before obtaining a chest radiograph. The presumptive diagnosis and decision to institute intercostal drainage were compared with the radiological diagnosis and the actual decision in each patient. A policy of selective drainage of large pleural collections was employed. The series consisted mainly of stab injuries in young men. Physical examination accurately diagnosed 13 of the 14 large pneumo- or haemothoraces. This reliability combined with the selective drainage policy showed that physical examination accurately predicted the need for tube thoracostomy with a sensitivity of 96 per cent and a specificity of 93 per cent. This study suggests that experienced clinicians should not hesitate to institute immediate life-saving intercostal drainage when needed, before a chest radiograph is obtained.