Protein kinase C Related Kinase 1 (PRK1) has been shown to be involved in the regulation of androgen receptor signaling and has been identified as a novel potential drug target for prostate cancer therapy. Since there is no PRK1 crystal structure available to date, multiple PRK1 homology models were generated in order to address the protein flexibility. An in-house library of compounds tested on PRK1 was docked into the ATP binding site of the generated models. In most cases a correct pose of the inhibitors could be identified by ensemble docking, while there is still a challenge of finding a reasonable scoring function that is able to rank compounds according to their biological activity. We estimated the binding free energy for our data set of structurally diverse PRK1 inhibitors using the MM-PB(GB)SA and QM/MM-GBSA methods. The obtained results demonstrate that a correlation between calculated binding free energies and experimental IC50 values was found to be usually higher than using docking scores. Furthermore, the developed approach was tested on a set of diverse PRK1 inhibitors taken from literature, which resulted in a significant correlation. The developed method is computationally inexpensive and can be applied as a postdocking filter in virtual screening as well as for optimization of PRK1 inhibitors in order to prioritize compounds for further biological characterization.