For metered dose inhalers (MDIs), high-flow cascade impaction with a United States Pharmacopia (USP) throat provides a useful prediction of in vivo lung and oropharyngeal aerosol deposition. Particles expected to deposit in the lung are included in the "fine particle fraction" measured on the bench. Comparable in vitro standards are not available for nebulizers. The present study compared aerosol deposition in an in vitro model using low-flow cascade impaction with deposition in vivo in human subjects. A low-flow (1 Lmin), 10-stage cascade impactor measured aerodynamic distributions of aerosolized interferon-gamma (IFN-gamma) from two nebulizers (Misty-Neb and AeroEclipse). (99m)Technetium diethylene triaminepenta-acetic acid ((99m)Tc-DTPA) was used as the radiolabel. Two bench conditions were specified: no breathing (standing cloud) and simulated ventilation with a piston pump (tidal volume 750 mL frequency 25 per minute and duty cycle 0.5). Mass median aerodynamic diameter (MMAD) for both nebulizers was affected by ventilation (Misty-Neb vs. AeroEclipse: 5.2 vs. 4.6 microm for standing cloud and 3.1 vs. 2.2 microm during ventilation). In three subjects, measured values of oropharyngeal deposition averaged 68.1 +/- 0.08% for Misty-Neb and 30.9 +/- 0.03% for AeroEclipse. In vivo deposition patterns compared to aerosol distributions from both nebulizers indicated that, for wet nebulization, penetration of aerosol beyond the upper airways (fine particle fraction) will occur only for aerosol particles below 2.5 microm. This assessment requires that the bench aerosol distribution be measured under conditions of clinical use (i.e., during tidal breathing).