Waste streams, especially food waste, present the advantage to be renewable sources of organic matter for the biological production of hydrogen by Dark Fermentation (DF). Nevertheless, substrates composition can impact hydrogen production and stability of the DF process. In particular, carbohydrate-rich substrates have a significant positive influence on hydrogen yield. However, waste can also be composed of other compounds such as lipids or proteins, whose effects on DF are poorly understood. Other parameters than macro-molecular composition (carbohydrates, lipids, proteins), such as biodegradability, could help to understand the variability of hydrogen production from different types of waste. Biodegradability is limited by the bio-accessibility and complexity of the organic compounds and can be estimated using these indicators. The study aimed to provide new insights into the correlations between macro-molecular composition, bio-accessibility, complexity and microbial hydrogen production, for individual and mixed substrates. The novelty of the present study relied on the identification of biodegradability indicators that can be linked to DF performances. The substrates studied corresponded to reconstituted biowaste (bread, French fries, meat, red berries, vegetables, breaded fish, yogurt), in association with cover energy crop (rye silage). Seven mixtures of these individual substrates were also investigated, with an initial composition that was modified by doubling or halving the raw percentage of a given component. The Biochemical Hydrogen Potential (BHP) and composition of individual substrates and mixtures were analyzed: soluble and available sugars (Anthrone method after mild hydrolysis), proteins (Total Kjeldahl Nitrogen) and lipids (Soxhlet extraction). The results showed that the substrates with the highest soluble sugar content presented the highest BHP, with 184.63 mL H2/gVS and 137.64 mL H2/gVS for bread and French fries, respectively. Statistical analyses (principal component analysis and matrix of Pearson correlations) were performed and confirmed the positive linear correlation between hydrogen yield and soluble carbohydrates (R² =0.7). Interestingly, BHP results for mixtures were higher (between 31 to 85 % higher) than theoretical ones, whose values were calculated from the weighted sum of the performances of individual components. This non-additivity could be explained by changes in biodegradability when working in mixture conditions. To confirm this hypothesis, bio-accessibility analyses based on a novel fractionation method of organic matter coupled with complexity analyses of organic matter (3D-fluorescence analysis) are in progress. The final results of this study will help to determine indicators for selecting appropriate substrates for DF, related to both their composition and structure.