Many paleoenvironmental studies based on benthic foraminiferal assemblages use different protocols for sample analysis. A standardized protocol has been recently established for biomonitoring applications, but for paleostudies, the influence of size fraction on benthic foraminiferal composition and biodiversity is poorly documented. We studied fossil foraminiferal assemblages along two paleorecords (BOBGEO-CS05 and SU81–44) from the Bay of Biscay covering the last ~35 ka cal BP. We investigated diversity and community composition to compare the impact of each size fraction (63-150 μm, >150 μm, >63 μm) on environmental interpretations. Foraminiferal diversity was affected by the accumulation of small opportunistic species. In terms of faunal composition, both paleorecords displayed a different pattern depending on the size fraction selected. While in both cores, the 63-150 μm fraction blurred the signal of some rare indicator species, our results show that i) in BOBGEO-CS05, it yielded no extra ecological information compared to the large fraction whereas ii) in SU81–44, it contained small opportunistic species that were not present in the >150 μm, impacting therefore paleoenvironmental interpretations. According to these findings, we recommend: i) to focus on the large fraction for a thorough taxonomic determination and a detailed analysis of benthic assemblages, and ii) to analyse the small fraction separately after a taxonomical identification of major species and strategic selection of studied samples. Although the 125 μm size limit was not tackled in this study, we recommend to use it for the limit between the small and large fractions instead of 150 μm for harmonization with the previously published standardized protocol for living faunas.