Background: Recent ICH guidances (Q8 and Q9) and FDA Guidances (PAT, Quality System, etc.) recognized the importance of using a Quality-by-Design (QbD) approach for pharmaceutical development, process and product design, and manufacturing improvement. Design of Experiment (DOE) as a key element of the QbD paradigm, if used appropriately, can be an effective tool to mitigate risk of producing pharmaceutical products of low quality. DOE is an efficient tool to develop process understanding and to construct a design space. This purpose of this study was to identify critical process variables and detect possible interactions between variables for a pharmaceutical capsule filling unit operation. Capsule filling process performance is dependent on capsule filling machine and the formulation. Variation in capsule weight clearly results in variation in drug content. Methods: The formulation consisted of 4% aspirin and 96% MCC. Two full factorial designs were created to include selected formulation [aspirin type (coarse and fine), filler MCC (AvicelŪ PH 200 and PH 301), lubricant level (0 and 0.5%)], and process [(capsule size (#0 and #1), filling machine type (Zanasi and H&H)] variables. A total of 24 batches were manufactured. Ten capsule samples were collected at each of the predefined time points throughout each production run. Particle size (Malvern Mastersizer), flowability (Shear Cell and a Flodex Tester), bulk density and tapped density (Scott Volumeter and the USP tapped density method) were measured. The coefficient of variation (CV) for capsule weight and aspirin content were determined for each interval sampled (CVi). Main effects and interactions were assessed by ANOVA. Principal component analysis (PCA) was used to analyze the variability in the powder's physical properties. Results: For average capsule fill weight, machine type was found to be the most significant variable. For percent aspirin content, capsule filling machine type, lubricant level, interactions between filler type/capsule size and lubricant level/capsule size were found to be significant. For CVi of aspirin content, only aspirin particle size was significant. For capsule weight average CVi, filler type, lubricant level, capsule filling machine type, and interaction between aspirin particle size/capsule filling machine type were found to be significant. Carr's index was found to be the main contributor to the first principal component (PC) in the dataset of the powder physical properties. Conclusions: Integration of DOE, multivariate statistical data analysis, and product/process characterization can be used to identify critical process/product variables and detect possible interactions between the formulation/process variables. The knowledge gained through this integrated QbD approach can provide scientific justifications for certain regulatory actions in the section of Chemistry, Manufacturing, and Control (CMC).