Thanks to the advent of clock synchronization via the global positioning system (GPS), phasor measurement units (PMUs) collecting both the magnitude and phase angle with higher precision have implications for monitoring performance across the wide area of smart grids. Owing to high PMUs installation costs, developing optimal strategy of PMUs placement is important to monitor the transmission system status in the wide area. To characterize the performance of line outage identification, this paper firstly proposes the statistical model to describe the average identification capability of multiple line outages. Using this model, we develop a global optimal PMUs placement strategy to maximize the average identification capability under the budget of PMUs. Furthermore, a greedy heuristic strategy is developed to bypass the combinatorial search to achieve a sub-optimal solution at a reduced complexity. The proposed methods are used to benchmark the optimal PMU placement solutions for the IEEE 14-and 57-bus systems. The experimental study demonstrates that the proposed techniques successfully select optimal PMU locations, targeting on maximum the average identification capability. The result shows that the proposed PMUs placement strategies improve about 10% location identification performance of multiple line outages when compared to random PMUs placement method.