Background: Recent FDA's Quality Initiatives (PAT, cGMP for the 21st Century, Quality System, and Critical Path) have highlighted the necessity and critical importance of building quality into the pharmaceutical products, for the benefit of the public. Quality can't be tested into products. However, the traditional release testing more falls into quality assurance domain and not necessarily build quality into products. Using PAT tools and QbD principles can advance our understanding of pharmaceutical development and manufacturing processes. Traditionally, the end-point for a powder blending process is time-based and determined empirically. The objective of this study was to develop an integrated QbD approach using process monitoring to determine the end-point and constituent concentrations of final blending mixture. Methods: A mixture design was created to include 26 powder formulations consisting of one API and three excipient components (A, B, C). For each powder formulation, about 5 grams in total was placed in a 20 ml scintillation vial for blending using a Turbula shaker-mixer. The mixer was stopped at various time points to enable NIR scan of the powder mixture and sampling for UV assay. Moving block standard deviation was applied to the NIR spectra for trend analysis and blending end-point determination. To quantify the concentrations of well-blended powder mixture, two types of multivariate calibration models (PLS/PCR??) were constructed and validated. The first model was constructed to correlate the UV spectral data with the targeted concentrations of a binary mixture (API and excipient component C). This model was then used to predict the concentrations of API and excipient C in the final well-blended powders. The second model was constructed to correlate the NIR spectra of well-blended 4-component powder mixture and the concentrations of blended powders. The prediction results of concentrations of API and excipient C from these two types of calibration models were compared. Results: The time to reach blending end-point is related to a number of factors such as the composition, the scale of scrutiny, and the homogeneity for both API and excipients in the formulation. The present study shows that the blending time end-point ranged from a few minutes to 45 minutes. The multivariate PLS and PCR models based on UV spectra has R2 value of 0.99 and could predict the concentrations of both API and excipient C well for binary powder mixture. The multivariate PLS and PCR models based on NIR spectra could predict the concentrations of both API and excipients well for the final well-blended 4-component mixtures. Conclusions: An integrated QbD approach was developed to determine the blending end-point and concentration quantification for a 4-component pharmaceutical powder blending system.