Introduction. Microfiltration is largely applied in the food sector for the separation and concentration of proteins. For example, the skimmed milk microfiltration is implemented in the industry to separate casein micelles (retentate) from serum proteins (filtrate) and to obtain high purity serum protein fractions before further processing (ultrafiltration and spray-drying). The efficiency of the process, and especially the serum protein transmission, depends on the accumulation of casein micelles at the membrane surface. To overcome this problem, milk microfiltration is operated with specific ceramic tubular membrane concepts: the Uniform Transmembrane Pressure (UTP) concept that makes it possible to get a low uniform transmembrane pressure (TMP) along the membrane with an additional pump on the permeate side; the Graded Permeability (GP) membrane concept that creates uniform permeate flux along the membrane with a membrane with higher hydraulic resistance at the entrance, when TMP is higher. Unfortunately, the performances of these ceramic membrane concepts have never been compared in a wide range of operating conditions. Experimental/methodology. This study proposes a comparison of the performances (fouling, selectivity) of the UTP and GP concepts in the case of milk microfiltration operating in feed and bleed mode, at 50°C in a large range of VRR, volume reduction ratio (1-3.5) and TMP (0.2 – 2 bar). Two protocols have been implemented with ceramic membrane Pall 0.1µm7P1940, UTP and GP configurations: - parametric study, which consists in step by step variation of TMP. - study as a function of time (4-5 hours) at constant VRR (recirculation of both retentate and permeate). Results and discussion. Despite the fact that both concepts used membranes with similar filtering layers, the performances of the two ceramic membrane concepts show large discrepancies, especially at high VRR (3.0-3.5) where fouling is more severe. The results are discussed considering the difference in membrane concepts features. The optimal performances are not always obtained under the industrially recommended conditions, which leaves room for significant improvements of existing filtration plants. Moreover, these results allow the definition of optimal operating conditions for each volume reduction factor , and thus each stage of an industrial skimmed milk plant using ceramic membranes