Background There is increasing use of therapeutic aerosols in patients undergoing mechanical ventilation. Few studies have measured aerosol delivery to the lungs under these conditions with adequate experimental methods. Hence this study was performed to measure pulmonary aerosol deposition and to determine the reproducibility of the method of measurement during mechanical ventilation. Methods Nine male patients were studied during mechanical ventilation after open heart surgery and two experiments were performed in each to determine the reproducibility of the method. A solution of technetium-99m labelled human serum albumin (Q""Tc HSA (50 ,ug); activity in experiment 1, 74 MBq; in experiment 2, 185 MBq) in 3 ml saline was administered with a Siemens Servo 945 nebuliser system (high setting) and a System 22 Acorn nebuliser unit. Pulmonary deposition was quantified by means of a gamma camera and corrections derived from lung phantom studies. Results Pulmonary aerosol deposition was completed in 22 (SD 4) minutes. Total pulmonary deposition (% nebuliser dose (SD)) was 2*2 (0.8)% with 1*5% and 0 7% depositing in the right and left lungs respectively; 0.90%o of the nebuliser activity was detected in the endotracheal tube or trachea and 51% was retained within the nebuliser unit. Considerable variability between subjects was found for total deposition (coefficient of variation (CV) 46%), but within subject reproducibility was good (CV 15%). Conclusions Administration of aerosol in this way is inefficient and further research is needed to find more effective alternatives in patients who require mechanical respiratory support. This method of measurement seems suitable for the assessment of new methods of aerosol delivery in these patients. (Thorax 1993;48:154-159)