The current model for hydrogen flow in the aldose-ketose isomerases is probably incorrect. Enzymes of this class are characterized by both hydrogen transfer and proton exchange in the interconversion of substrate and product. The transfer is believed to be due to the action of a unique basic residue in the active site. Exchange is presumed to occur by dissociation of the abstracted proton and reassociation from the medium prior to its transfer to the intermediate enediol on the way to product. Dissociation of a necessary proton from the intermediate state imposes limits on the overall catalytic rate depending on the pKa of the protonated base and the pH of the medium. A case in point is triose-P isomerase (TIM), where kcat is approximately 10(4) s-1. T-Labeled substrate is found to lose approximately 95% of its T to the medium when totally converted to product. Although the active site base is believed to be a glutamate of pKa = 3.9, the pH dependence of maximum velocity is known to be flat up to pH 10. The loss of hydrogen required to form product as indicated by isotope exchange must be restored completely at this high pH, requiring a base of very high pKa, or there must be some other explanation for the loss of isotope. The present study demonstrates the existence of a single proton on human and rabbit TIM and three protons on yeast TIM that rapidly exchange with the abstracted proton at the E.enediol state internal exchange. Exchange with the medium external exchange occurs from the enzyme after substrate or product has dissociated.(ABSTRACT TRUNCATED AT 250 WORDS)