This thesis proposes a method for designing global Power System Stabilizers (PSS) using Phasor Measurement Unit (PMU) data. The method first designs local PSS using existing input/output selection techniques, but finds that the number of selected inputs and outputs is too large. The thesis then suggests a new algorithm that uses bus/generator clustering with independent component analysis and principal component analysis to reduce the number of input/output candidates. Finally, the design of global PSS is performed using the optimized input/output selection from the new method, PMU data, a genetic algorithm optimization method, and linear matrix inequalities.