This presentation will summarize ongoing research on risk and reliability-based decision methods for homeland security applications at Vanderbilt University, within the NSF-IGERT multi-disciplinary graduate program on Risk and Reliability Engineering and Management. Risk-based decision models are being developed at two levels, local and global, respectively: (1) protection systems for individual facilities against adversary attacks, and (2) system of systems decision-making for multiple, interdependent infrastructures. The local facility protection systems research is in the context of stochastic network interdiction, and incorporates two elegant methods of model-based reliability analysis, and reliability-based optimization. The research extends current network interdiction methods to include multiple safeguard types, and comprehensive consideration of uncertainties in safeguard and detector performance, response time, etc. The global decision support system builds on the network optimization methodology to facilitate resource allocation decisions, risk communication, and risk management for multiple, interdependent infrastructures facing the risk of terrorist attack, using multi-disciplinary optimization concepts.