The aim of this work was to investigate mixing and liquid-to-gas mass transfer of hydrogen in relation to hydrogen production in the dark fermentation process as a function of agitation conditions and digestate viscosity. Experiments were carried out in a baffled mechanically-stirred reactor equipped with a dual-stage impeller using five levels of viscosity. Biohydrogen production was studied using glucose as substrate under controlled pH. Three experimental techniques, namely local conductimetry, chemical decolorization and Planar Laser Induced Fluorescence were used to measure mixing time t(m) and describe the flow pattern. The effects of inter-impeller clearance and tracer injection position were also studied. Then, (k(L)a)(H2) was deduced from dynamic deaeration/aeration experiments. Experimental results showed that biohydrogen production presented a maximum in the transitional flow regime (Re about 200), and fell under turbulent flow (Re > 1000). Similarly, the evolution of (k(L)a)(H2) was better described by Re than by the volumetric power input, contrary to literature data. Finally, the Damkohler number showed that hydrogen production was limited by liquid-to-gas mass transfer in the laminar regime and that maximum reaction rate could be reached only due to dissolved H-2 supersaturation in the liquid phase. Conversely, the steep decrease of H-2 production under turbulence conditions could be attributed neither to mass transfer, nor to mixing conditions, highlighting a probable negative interaction between turbulent eddies and biomass aggregates. Regarding k(L)a center dot t(m), the transitional flow region also approached ideal mixing, which strengthened the conclusion that H-2 production was optimized in the transition region in the dark fermentation process.