The transfer of tritium between adenosylcobalamin and substrate in the reaction catalyzed by glutamate mutase was examined to investigate the possibility of a protein-based radical intermediate. There was no evidence that tritium was transferred to the protein during the reaction, as tritium neither became stably bound to the protein nor exchanged with water. The kinetics of tritium transfer from adenosylcobalamin to 3-methylaspartate was investigated. Both the transfer of tritium to product and the exchange of enzyme-bound and free coenzyme contribute to the kinetics of tritium loss from adenosylcobalamin. By varying the experimental conditions, the rates of both coenzyme exchange and tritium transfer could be measured. Exchange of adenosylcobalamin with enzyme is very slow, k off = 0.01 s-1, which may reflect a conformational change in the coenzyme and/or protein involved in forming active holo enzyme. The rate constants for the loss of tritium from adenosylcobalamin and the appearance of tritium in 3-methylaspartate are much faster and very similar, k = 0.67 +/- 0.05 s-1 and k = 0.50 +/- 0.05 s-1, respectively, consistent with the transfer of tritium occurring directly between coenzyme and substrate. The isotope effect, calculated from the rate constants for tritium transfer, and kcat, determined for the overall reaction under the same conditions, are between 13.5 and 18. These values are typical of primary isotope effects seen for enzymes in which hydrogen transfer is substantially rate limiting. A protein radical, therefore, appears unlikely to feature in the mechanism of this enzyme.