An uncharacterized minor transient product, observed in our earlier studies of substrate turnover by the E48Q mutant of Rhodospirillum rubrum ribulose-bisphosphate carboxylase/oxygenase (Lee, E. H., Harpel, M. R., Chen, Y.-R., and Hartman, F. C. (1993) J. Biol. Chem. 268, 26583-26591), becomes a major product when it is trapped and stabilized with borate as an additive to the reaction mixture. Chemical characterization establishes this novel product as D-glycero-2,3-pentodiulose 1,5-bisphosphate, thereby demonstrating oxidation of the C-3 hydroxyl of D-ribulose 1,5-bisphosphate to a carbonyl. As the formation of the novel oxidation product is oxygen-dependent and generates hydrogen peroxide, its precursor must be a peroxy derivative of ribulose bisphosphate. Thus, discovery of the dicarbonyl bisphosphate lends direct support to the long standing, but heretofore unproven, postulate that the normal pathway for oxidative cleavage of ribulose bisphosphate by the wild-type enzyme entails a peroxy intermediate. Our results also suggest that stabilization of the peroxy intermediate by the wild-type enzyme promotes carbon-carbon scission as opposed to elimination of hydrogen peroxide.