The quantitative measurement of regional aerosol deposition in human lungs using two-dimensional (2D) gamma scintigraphy has proven to be useful in therapeutic and diagnostic aerosol studies. The penetration index (PI) has been defined as the ratio of activity in a peripheral lung zone to a central lung zone, but the ability to discriminate between aerosol deposition in the large airways and lung parenchyma is reduced by the fact that the latter overlies the former in the central zone. To overcome this, we used a three-dimensional (3D) technique. Seven healthy subjects inhaled isotonic saline aerosols containing 99mTc-DTPA on two occasions. The droplets had a mass median aerodynamic diameter (MMAD) of either 2.6 or 5.5 microns (with geometric standard deviations [sigma g] of 1.4 and 1.7, respectively). Transmission tomography was performed on each subject to delineate lung boundaries in 2D and 3D. After inhalation, anterior (A) and posterior (P) images were collected and a tomographic study performed. Mid-lung slices were taken from coronal (CC) and transverse (TC) sections. PI was calculated on the 2D images (AP and P) and the 3D slices (CC and TC) using exactly defined regions. The PI values were smaller for the large droplet aerosol (5.5 microns) in all subjects and methods. The relative differences in PI between large and small (2.6 microns) droplet studies (d values) were greater and less variable for the 3D methods (TC, 56.5 +/- 11.4% and CC, 52.4 +/- 12.3%) compared to the 2D methods (P, 25.4 +/- 17.1% and AP, 38.3 +/- 15%; p less than 0.005). We found the 3D methods to be more sensitive for discriminating between aerosol deposition in large and small airways than were the conventional 2D methods.