In designing power system stabilizer (PSS), a major difficulty is to successfully treat the system's uncertainties. These uncertainties arise because of changes in operating conditions, approximations in modeling, parameter variations caused by faults, etc. It is known that in the presence of uncertainties, conventional methodologies such as linear optimal technique, adaptive controls, etc., may fail to guarantee the stability of the system. On the other hand, the H" control theory provides potential ability to overcome this problem.However, some limitations still exist in the treatment of uncertainty. That is, the existing standard H"-PSSs cannot adequately treat the system uncertainties. Moreover, performance problems can arise in this approach due to the pole-zero cancellation phenomenon.To deal with the above mentioned limitations, a new design methodology for H" PSS based on the 'numerator-denominator' uncertainty representation is proposed, where partial pole placement technique is used. Simulation results suggest that the proposed PSS is more robust and less sensitive to disturbances than both the conventional PSS and the standard H,-PSS. The superiority of the proposed PSS has also been confirmed by computational results of stability margin as well as the Bode plots of the sensitivity functions.