The production of hydrogen from substrates and substrate mixture of sugar alcohols and Polydextrose was determined, both in vivo using the breath hydrogen test, and in vitro, using human faecal microorganisms in anaerobic culture. One objective was to test a previous assumption that the stoichiometry of hydrogen production from different alternative carbohydrates is similar. Another objective was to discover whether hydrogen responses from mixtures of substrates were simply additive, or whether interactions occurred. The breath tests were performed in a 10 subject x 10 substrate factorial design with substrates and substrate mixtures (5-11 g) administered in 42 g chocolate confectionery. Incorporation of the alternative carbohydrates lactitol (L), Isomalt (I) and Polydextrose (P) into otherwise conventional confectionery increased breath hydrogen production by approximately 112, 73 and 11%/g respectively. There was no interaction between L and I or between P and I, but a combination of L and P approximately doubled the breath hydrogen anticipated from their individual contributions (P < 0.05). Anaerobic cultures showed a sixfold range in the efficiency of converting individual substrates and mixtures to hydrogen gas (0.003-0.018 kJ H2 per kJ carbohydrate). The positive interaction between L and P, and the lack of interaction between L and I, and between P and I, found in vivo were reproduced in vitro. The work showed that interpretation of the hydrogen breath test is confounded by differing stoichiometries for hydrogen production, by interaction between substrates and by an uncertain extent to which small intestinal hydrolysis yielding species with a fermentation stoichiometry that differs from the parent substrate.