This paper describes a risk analysis approach for assessing the progress of another country's nuclear weapons program over time. To handle the dynamics inherent to nuclear weapons development, we embed a semi-Markov decision process into a Bayesian network. The Bayesian network accounts for distributions on the time to transition between possible states of the nuclear weapons program. Our approach enables analysis of the country's nuclear weapons program decisions by identifying how each decision maker would direct the program given the domestic, international, and security influences affecting the country. We demonstrate the model with a case study of Pakistan's nuclear weapons program.