Pulmonary toxicity developed in 15 (17%) of 89 patients treated with amiodarone during a follow-up period of 2 weeks to 54 (mean 20 +/- 15) months. Prospective evaluation of serial pulmonary function tests in 67 patients demonstrated both a significant decrease from baseline in three of six variables in patients with toxicity at the time of diagnosis and a significant difference compared with the same variables in patients without toxicity. The most significant of these was the diffusing capacity for carbon monoxide (DLCO). An individual decrease in DLCO greater than or equal to 15% gave an optimal sensitivity of 100% and a specificity of 89% for the diagnosis of pulmonary toxicity. However, a decrease in DLCO greater than or equal to 15% did not alone warrant a change in therapy in asymptomatic patients. Although higher maintenance doses of amiodarone appeared to be related to the development of this complication, an abnormal baseline DLCO (less than 60% of predicted) with or without an initial abnormal chest roentgenogram did not predispose to pulmonary toxicity.