Powder flow of mixtures is complex and not properly understood. The selection of drug-excipient blends with inadequate powder flow can lead to quality issues of the final dosage form. Therefore, this work aims at a better understanding of how changes in powder flow of binary blends can lead to weight variability in pharmaceutical capsule filling. We used image-analysis-based powder avalanching and shear cell testing to study blends of paracetamol and microcrystalline cellulose. A pilot-scale machine with dosator principle was employed for encapsulation. As a result, the powder flow properties improved generally with rising amounts of microcrystalline cellulose. However, a negative correlation was observed between avalanche angle and angle of internal friction. Results were discussed and percolation theory was considered to explain abrupt changes in the observed flow properties. This was particularly helpful for analysis of the capsule-filling data, since capsule weight variability displayed a threshold behavior as a function of the mixture fraction. The capsule weight variability correlated with the angle of internal friction as well as with the angle and the energy of avalanches. Based on the results we proposed a strategy of how to design minimal weight variability into powder-filled capsules.