This thesis proposes an improved procedure to watermark Intellectual Property Cores at Register Transfer Level using Genetic Algorithms. First, watermarking signature and Intellectual Property Core's behavioral description are translated into Finite State Machines in preparation for merging. The resulting Finite State Machine contains the watermarked Intellectual Property Core maintaining its original functionality without disruption. Next, a reduction procedure is applied to the watermarked design. At this stage, dealing with hanging states is challenging, if any of these is deleted, the watermark could be removed and possibly the original Intellectual Property Core functionality would not be disrupted. Both Finite State Machine merging and reduction are NP-Complete problems. In this study an improved objective function is proposed to accurately model the Finite State Machine reduction problem while applying Genetic Algorithms as optimization techniques at both stages. Empirical results show a significant improvement in terms of the number of final hanging states and watermark embedding strength as regards previous reported approaches. Results of applying the proposed technique to watermark a number of Finite State Machines are presented and discussed.