Screw feeders are integrated as dispensing units in most continuous manufacturing platforms. Hence, characterizing and modelling the residence time distribution (RTD) of materials in feeders is indispensable to understand the traceability of raw materials from the drum till tablet, enabling the separation of non-confirming material. The proposed methodology addressed this leap in knowledge by characterizing materials, performing RTD trials according to an experimental design, applying RTD models and establishing a partial least square (PLS) regression model that links the material properties and process variables with the RTD responses as outputs. Results showed that RTD in screw feeders can be represented by a combination of plug-flow and mixed-flow. Three variables were found to impact the residence time distribution in feeders: flow rate, hopper level and conditioned bulk density. Interestingly, the plug-flow fraction was not affected by variation in flow rate or material properties. Consequently, simple PLS models could be developed that use density and flow rate to predict RTD at a given hopper level. This approach is powerful for RTD prediction based on bulk density in the early phases of development when control strategies for clinical manufacturing need to be established and material availability is still limited.