Quantitative macro-Raman spectroscopy was applied to the analysis of the bulk composition of pharmaceutical drug powders. Powders were extracted from seven commercial lactose-carrier-based dry-powder inhalers: Flixotide 50, 100, 250, and 500 μg/dose (four concentrations of fluticasone propionate) and Seretide 100, 250, and 500 μg/dose (three concentrations of fluticasone propionate, each with 50 μg/dose salmeterol xinafoate ). Also, a carrier-free pressurized metered-dose inhaler of the same combination product, Seretide 50 (50 μg fluticasone propionate and 25 μg salmeterol xinafoate per dose) was tested. The applicability of a custom-designed dispersive macro-Raman instrument with a large sample volume of 0.16 μL was tested to determine the composition of the multicomponent powder samples. To quantify the error caused by sample heterogeneity, a Monte Carlo model was developed to predict the minimum sample volume required for representative sampling of potentially heterogeneous samples at the microscopic level, characterized by different particle-size distributions and compositions. Typical carrier-free respirable powder samples required a minimum sample volume on the order of 10(-4) μL to achieve representative sampling with less than 3% relative error. In contrast, dosage forms containing non-respirable carriers (e.g., lactose) required a sample volume on the order of 0.1 μL for representative measurements. Error analysis of the experimental results showed good agreement with the error predicted by the simulation.