Phasor measurement units (PMUs) have offered significant advancements in power system operation and control. Despite these advantages, the industry has been slow in adopting PMU technology. This slow adoption is partially due to the high cost of installing PMUs, which motivates the optimal PMU placement (OPP) problem. The work presented in this paper is motivated by the idea that the value of a high cost asset can be increased by deploying it to improve network fault-tolerance. This strategy of deploying PMUs in the proximity of higher probability contingencies increases the likelihood of more effective remedial actions, both preventive and corrective. The proposed OPP therefore considers both the cost and the extent to which the PMU placements will benefit system observability in the presence of network vulnerabilities. A bi-level framework is used for optimizing the network vulnerability and the installation cost of PMUs. In the proposed framework, the installation cost of PMUs is treated as the primary objective, and the observability, which incorporates the vulnerability analysis, is treated as the secondary objective. The proposed approach can be used as a multistage installation process or as a single-stage installation process. This approach uses evolutionary algorithms to optimize and prioritize locations and installation cost of the PMUs. The proposed approach is tested on the IEEE 14-bus, IEEE 30-bus, and IEEE reliability test system.