Crystallographic studies of D-xylose isomerase (D-xylose ketol-isomerase, EC 5.3.1.5) incubated to equilibrium with substrate/product mixtures of xylose and xylulose show electron density for a bound intermediate. The accumulation of this bound intermediate shows that the mechanism is a non-Michaelis type. Carrell et al. [Carrell, H. L., Glusker, J. P., Burger, V., Manfre, F., Tritsch, D. & Biellmann, J.-F. (1989) Proc. Natl. Acad. Sci. USA 86, 4440-4444] and the present authors studied crystals of the enzyme-substrate complex under different conditions and made different interpretations of the substrate density, leading to different conclusions about the enzyme mechanism. All authors agree that the bound intermediate of the sugar is in an open-chain form. It is suggested that the higher-temperature study of Carrell et al. may have produced an equilibrium of multiple states, whose density fits poorly to the open-chain substrate, and led to incorrect interpretation. The two groups also bound different closed-ring sugar analogues to the enzyme, but these analogues bind differently. A possible explanation consistent with all the data is that the enzyme operates by a hydride shift mechanism.