The objectives of this study were to develop a predictive statistical model for low-fill-weight capsule filling of inhalation products with dosator nozzles via the quality by design (QbD) approach and based on that, to study the effect of scaling up a capsule filling operation of capsule fill weight (1- 45mg) and weight variability (RSD) of typical inhalation carriers. Various controllable process parameters and uncontrolled material attributes of 12 powders were initially screened using a linear model with partial least square (PLS) regression to determine their effect on the critical quality attributes (CQA) (fill weight and weight variability). After identifying critical material attributes (CMAs) and critical process parameters (CPPs) that influenced the CQA, model refinement was performed to study if interactions or quadratic terms influence the model. Based on that, we developed a linear predictive model for fill weight and a model that provides a good approximation of the fill weight variability for each powder group. We validated the model, established a design space for the performance of different types of inhalation grade lactose on low-fill weight capsule filling and successfully used the CMAs and CPPs to predict fill weight of powders that were not included in the development set on lab scale. Finally, the scale up performance was tested for a variety of capsule filling speeds, pre-compression ratios and nozzle diameters. The Design Space developed with the Labby (lab-scale machine) can be safely used to set the process parameters to obtain a specific fill weight in the industrial scale machine.