The particle sizing performance of a Next Generation Pharmaceutical Impactor (NGI) was compared to that of an Andersen cascade impactor (ACI). A single lot of Vanceril® MDIs containing beclomethasone dipropionate (BDP) was used throughout. MDIs were sampled into NGI and ACI in accordance with USP recommendations, at 30.0 and 28.3 L/min, respectively, following 1, 2, 6, and 30 actuations with or without a silicone cup or stage coating, to determine the apparent particle size distributions (PSD) of BDP. The mass balance and the statistical comparability of drug deposits were assured on a “per actuation basis” across all experiments, demonstrating “good cascade impactor practices.” Interstage deposition or “wall losses” in NGI were found to be lower than those in ACI, although their determination was laborious in NGI. The PSD profiles for Vanceril® from a single actuation were distinguishable between NGI and ACI, when uncoated collection surfaces were used, most specifically for drug mass <4‐μm aerodynamic diameter (p < 0.05). Silicone coating of collection surfaces and an increased number of actuations were shown to result in PSD profile shifts for both NGI and ACI. Such effects were most pronounced for NGI, although coating the collection surfaces and/or increasing the number of actuations improved drug retention significantly on the upper stages of NGI, and thereby, minimized the effects of particle bounce of BDP from Vanceril® MDIs. PSD profiles from a single actuation could be determined reliably in either of these impactors, provided that coated collection surfaces were employed; also, cumulative % mass undersize profiles were similar between instruments. However, small differences in PSD profiles still existed to support NGI's design claims for reduced “overlap” in its stage collection efficiency curves. © 2004 Wiley‐Liss, Inc. and the American Pharmacists Association J Pharm Sci 93:1828–1837, 2004