HpcG catalyses the hydration of a carbon-carbon double bond without the aid of any cofactor other than a simple divalent metal ion such as Mg(2+). Since the substrate has a nearby carbonyl group, it is believed that it first isomerises to form a pair of conjugated double bonds in the enol tautomer before Michael addition of water. Previous chemical studies of the reaction, and that of the related enzyme MhpD, have failed to provide a clear picture of the mechanism. The substrate itself is unstable, preventing co-crystallisation or soaking of crystals, but oxalate is a strong competitive inhibitor. We have solved the crystal structure of the protein in the apo form, and with magnesium and oxalate bound. Modelling substrate into the active site suggests the attacking water molecule is not part of the metal coordination shell, in contrast to a previous proposal. Our model suggests that geometrically strained cis isomer intermediates do not lie on the reaction pathway, and that separate groups are involved in the isomerisation and hydration steps.