Biochemical decompression has been proposed as a method for reducing the amount of time required for deep-sea divers to return to the surface. Divers breathing H2/O2 mixtures would be presented with hydrogenase enzyme, and decompression would be accelerated by means of the enzymic removal of excess H2 from the tissues. We have studied FAD as a hydrogenase electron acceptor that is capable of transferring electrons derived from H2 oxidation directly to O2. Kinetic activity constants for the soluble hydrogenase from the bacterium Alcaligenes eutrophus H16 were determined with FAD, FMN and riboflavin as electron acceptors, and these values were compared with those obtained with the physiological electron acceptor NAD+. The Michaelis constants (K(m)) were similar for FAD, FMN and NAD. However, the maximal catalytic-centre activity (Kcat) was much lower for the flavins, and the catalytic efficiency (Kcat/K(m)) with FAD was 1/20th the value for NAD+. After enzyme-catalysed FAD reduction to FADH2, the FAD could be regenerated by addition of O2 and reduced again by the enzyme in the presence of H2. Thus FAD served as a regenerable electron shuttle between H2 and O2. H2O2, a by-product of FADH2 oxidation by O2, inhibited the enzyme. Much greater inhibition was observed with the reduced form of the enzyme. Active hydrogenase was efficiently encapsulated into human and pig red blood cells. Hydrogen consumption was seen with lysed carrier cells, but was demonstrated with unlysed carrier cells only when FAD was co-encapsulated along with enzyme. These results demonstrate that red blood cells encapsulating hydrogenase and FAD act as a system for continuous H2 consumption in a mammalian tissue without addition of exogenous factors, and such cells may provide a biotherapeutic method for reducing the risk and treatment of decompression sickness.