Molecular docking of substrates is more challenging compared to inhibitors as the reaction mechanism has to be considered. This becomes more pronounced for zinc-dependent enzymes since the coordination state of the catalytic zinc ion is of greater importance. In order to develop a predictive substrate docking protocol, we have performed molecular docking studies of diketone substrates using the catalytic state of carbonyl reductase 2 from Candida parapsilosis (CPCR2). Different docking protocols using two docking methods (AutoDock Vina and AutoDock4.2) with two different sets of atomic charges (AM1-BCC and HF-RESP) for catalytic zinc environment and substrates as well as two sets of vdW parameters for zinc ion were examined. We have selected the catalytic binding pose of each substrate by applying mechanism based distance criteria. To compare the performance of the docking protocols, the correlation plots for the binding energies of these catalytic poses were obtained against experimental Vmax values of the 11 diketone substrates for CPCR2. The best correlation of 0.73 was achieved with AutoDock4.2 while treating catalytic zinc ion in optimized non-bonded (NBopt) state with ?1.01 charge on the zinc ion, compared to 0.36 in non-bonded (?2.00 charge on the zinc ion) state. These results indicate the importance of catalytic constraints and charge parameterization of catalytic zinc environment for the prediction of substrate activity in zinc-dependent enzymes by molecular docking. The developed predictive docking protocol described here is in principle generally applicable for the efficient in silico substrate spectra characterization of zinc-dependent ADH.