In food industries, microfiltration is widely used for the separation of particles and/or microorganisms in many beverages as fruit juices. However, like in many other applications, an accurate prediction of juices filterability is not yet validated; filtration tests in lab-scale dead-end filtration units have been challenged by frequent deficiencies to describe filtration performance at large-scale. The aim of this work was to develop a suitable lab-scale filtration strategy in order to anticipate membrane fouling during crossflow microfiltration of orange juices, through a relevant evaluation of their fouling behavior. The impact of membrane average pore size, juice particles size distribution and shear forces on filtration tests in dead-end filtration cell was explored through a D-optimal experimental design. Microfiltration experiments at pilot-scale were carried out in order to demonstrate the relevance of the proposed strategy. Results showed that an anticipation of juice fouling propensities, at pilot-scale, was possible when stirred dead-end filtration test was carried out under the following conditions: ceramic membrane with an average pore size of 0.2 lm, juice fractionation prior to filtration (500 g/1 min) and rotational speed of the blade of 1000 rpm. These conditions, notably the ones related to the particle back-transport and fouling mechanisms, showed that dead-end filtration cell can be used as a process-specific anticipation tool to predict membrane fouling during fruit juices pilot-scale microfiltration.