<P><B>Abstract</B></P> <P>Batch tests of H<SUB>2</SUB> production from glucose, molasses and cheese whey by suspended and immobilized cells of <I>Thermotoga neapolitana</I> were conducted to develop a kinetic model of the process. H<SUB>2</SUB> production was inhibited by neither H<SUB>2</SUB> (up to 0.7mgL<SUP>−1</SUP>) nor O<SUB>2</SUB> (up to 0.2mgL<SUP>−1</SUP>). The H<SUB>2</SUB> specific rates obtained at different substrate concentrations were successfully interpolated with Andrew’s inhibition model. With glucose and molasses, biofilms performed better than suspended cells. The suspended-cell process was successfully scaled-up to a 19-L bioreactor. Assays co-fed with molasses and cheese whey led to higher H<SUB>2</SUB> productivities and H<SUB>2</SUB>/substrate yields than the single-substrate tests. The simulation of the suspended-cell continuous-flow process indicated the potential attainment of H<SUB>2</SUB> productivities higher than those of the batch tests (up to 3.6mmol<SUB>H2</SUB> h<SUP>−1</SUP> L<SUP>−1</SUP> for molasses and 0.67mmol<SUB>H2</SUB> h<SUP>−1</SUP> L<SUP>−1</SUP> for cheese whey) and allowed the identification of the optimal dilution rate.</P> <P><B>Highlights</B></P> <P> <UL> <LI> No H<SUB>2</SUB> or O<SUB>2</SUB> inhibition on H<SUB>2</SUB> production by <I>Thermotoga neapolitana</I>. </LI> <LI> Significant substrate (glucose, molasses, cheese whey) inhibition on H<SUB>2</SUB> production. </LI> <LI> With glucose and molasses, biofilms performed better than suspended cells. </LI> <LI> Scale-up of the suspended-cell process to a 19-L fermentor. </LI> <LI> Simulation of the continuous-flow process to evaluate the optimal dilution rate. </LI> </UL> </P>