The nucleoside phosphotransferase from malt sprouts contains one Mg2 per dimeric enzyme molecule. This cation can be removed by EDTA, while other chelating agents are less efficient. The metal-free apoenzyme is inactive. Activity can be restored to its initial value by Mg2 or Co2 and to a minor extent by Mn2, Zn2, Cu2, Ni2 and Fe2. Thermal stability is reduced when the metal is removed but can be restored by addition of Mg2. Adenosine 3',5'-phosphate (cAMP) and arsenate, strong competitive inhibitors, reduce the rate of inactivation by EDTA considerably but do not reduce the rate for the reconstitution with Mg2. We therefore conclude that the phosphate group interacts electrostatically with Mg2 and that the inhibitor is not bound to the apoenzyme. The Sp-isomer of adenosine 3',5'-thionophosphate is a hundred times stronger inhibitor than the Rp-isomer and ten times stronger than cAMP; this allowed us to determine the relative position of the Mg2 in the active site. The imidazolium cation, previously detected as an essential part of the active site, obviously forms a salt bridge to the carboxylate group which attacks the phosphorus opposite to the leaving alcohol group. This conclusion is derived from the fact that organic solvents increase the rate of formation of the phospho-intermediate considerably, while higher concentrations of salt decrease it strongly. In addition, the imidazolium cation seems to polarize the P = O bond and to stabilize the negative charge at the phosphoryl oxygen in the pentacoordinate transition state; this is followed from the pH-dependence of the hydrolysis reaction. The kinetic results reveal that Km does not represent a binding equilibrium, but is the ratio of the rates for decay and formation of the covalent intermediate, while kcat/Km is an indicator for the formation step of the acyl phosphate. On the basis of all these informations it should be allowed to formulate a reaction mechanism, in which all steps of the transferase reaction have to be microscopically reversible.