SUMMARYIn next generation bio-based refineries, hydrolysis and primary (or extractive) fermentations by undefined microbial cultures (UMC) are precursors of secondary bio-transformations, in which H 2 , CO 2 and mixed carboxylates are used as substrate for achieving added-value target products (e.g. bio-based chemicals, bio-plastics and pigments). Dark fermentation (DF) is the most simple UMC-driven hydrolysis and primary fermentations to extract gaseous and soluble mixtures of compounds from raw biomass. Which solid fractions (types of macro-molecules) of mixed raw organic matter (OM) are efficiently hydrolyzed + fermented during DF is an aspect that was rarely considered in depth.Here, a first attempt was made to propose a new approach for understanding the effects of DF on different fractions of biomass. A set of seven different biomasses underwent optimized DF tests and, for simplicity, only the gaseous main product, i.e. bio-hydrogen potential (BHP) production, was used as parameter to assess DF efficacy. BHP was studied in relation with OM characteristics: on one side, chemical composition (macro-molecular fractions) and, on the other side, bioavailability to UMC attack (using two different biological assays). BHP was found significantly correlated (Pearson's test for p < 0.05, n = 7) only to acid detergent lignin (negatively), soluble sugars and sugars + starch (positively). Bioavailability was negatively correlated with fibrous fractions and to fat-like fractions, but correlations with BHP were poorer (p > 0.05, n = 7). A statistical model (partial least square regression) was proposed for predicting BHP from OM characteristics, with interesting predictability. In the next future, the proposed approach should be widened to better understand the DF effectiveness not only referred to its gaseous products, but especially focusing on the wide range of soluble products (carboxylates), thought as substrates for secondary biorefinery.