1. This study was performed to determine the variability of two different scintigraphic methods of measuring pulmonary aerosol deposition, and to examine nebulizer particle size and drug output as potential sources of this variability. 2. A radioaerosol was produced from a 3 ml solution of 99m Tc-labelled colloidal human serum albumin (0.05 mg, 37 MBq) using a standard jet nebulizer and air compressor. This was inhaled on three separate occasions by nine healthy male subjects. On one of these occasions, a further inhalation was performed to assess immediate repeatability using increased 99m Tc activity (92 MBq). 3. Intrapulmonary aerosol deposition was measured with a γ-camera and was corrected for tissue attenuation and geometric distribution by using two different methods. 4. Estimated mean pulmonary deposition was 4.3% of the nebulizer dose using a lung phantom correction method, and 6.1% using a tissue attenuation method. For these two methods respectively variability between subjects (coefficient of variation) was 54 and 47%. For both methods, within-subject variability (coefficient of variation) was 37% between occasions and 23% within occasions. 5. The particle-size output of several nebulizers was highly reproducible (coefficient of variation < 4%), but the nebulizer mass and radionuclide output of two nebulizers was more variable (coefficient of variation 5–19%), and appeared to be an important contributor to the variability in pulmonary aerosol deposition. 6. The data presented here for pulmonary deposition, used with appropriate power statistics formulae, can be used to estimate the sample sizes required for comparative studies of lung aerosol deposition.