Résumé : The filtration method (FM) is the most effective isolation technique for Epsilobacteriaceae from stool samples. FM’s different adaptations make it difficult to compare data between studies. This study was performed in three phases to optimize FM from a routine laboratory perspective. In July–September 2014 (part I), FM was performed on Mueller–Hinton agar containing 5% sheep blood and Columbia agar containing 5% sheep blood. In July 2016 (part II), FM was performed using 0.60-μm pore size polycarbonate filters (0.6-PC filter) and 0.45-μm pore size cellulose acetate filters (0.45-AC filter); in January 2018 (part III), the addition of hydrogen to incubators was studied. On 1146 stools analyzed in part I, the positive samples that showed no growth on the Butzler medium (n = 22/72, 30.6%) had improved growth of Epsilobacteriaceae when using the Columbia instead of the Mueller–Hinton medium (21/22 strains vs. 11/22, p < 0.05). In part II, on 718 stools, 91 strains grew with FM (12.7%), more with 0.6-PC filter (90/91) than with 0.45-AC filter (44/91) (p < 0.05). In part III, 578 stools were cultured, 98 Epsilobacteriaceae strains grew with FM, and 7% hydrogen finding significantly more Epsilobacteriaceae than without hydrogen (90/98, 91.8%, vs. 72/98, 73.5%; p < 0.05). The use of a Columbia medium containing 5% sheep blood with 0.6-PC filters incubated at 37 °C in a 7% hydrogen-enriched atmosphere led to an almost fourfold increase in the isolation rate of Epsilobacteriaceae among the studied combinations. Reference centers for Campylobacter should use standardized protocols to enable the comparison of prevalence in space and time.