We compared the clinical performance of a new 99Tcm DTPA aerosol delivery system with steady-state 81Krm gas in the diagnosis of pulmonary embolic disease. Forty patients had aerosol and 81Krm ventilation imaging followed by perfusion scintigraphy. The combined aerosol/perfusion images were compared with 81Krm/perfusion images for diagnostic equivalence. Aerosol image quality was also compared directly with 81Krm. Three-quarters of the patients who had segmental ventilation-perfusion (V/Q) mismatch on 81Krm/perfusion images also had mismatch on aerosol/perfusion images. Of those who had no mismatch on Kr/perfusion images, it was possible to rule out mismatch on aerosol/perfusion images in 80% of cases. Fifteen per cent of aerosol images were uninterpretable due to excess deposition in large airways. This problem was significantly more frequent in smokers than in non-smokers. The average delivery efficiency of the system was 2%, in terms of the proportion of loaded activity retained in the lungs. We conclude that the new device gives accurate clinical information in most cases, but is of less value for smokers.