Nuclear deterrence theory in its many forms arose as a theoretical architecture with the goal of preventing rather than winning a nuclear war. Although evidence exists that nuclear weapons do deter full out war between nuclear armed rivals, the extent of this deterrent capability is much less clear. This paper analyzes the uncertainty of nuclear deterrence from the standpoint of both classic theoretical arguments and more recent empirical attempts. From both qualitative and quantitative perspectives, this paper finds cause to question the certainty that nuclear deterrence will inevitably hold in the future. Although nuclear war between nuclear rivals has never occurred, this lack of data is largely what makes predicting the continued success of nuclear deterrence in the future so difficult. In fact, from a certain probabilistic point of view, historical empirical evidence is not inconsistent from nuclear war between nuclear rivals being an event that occurs on average once every 100 years. Finally, this paper offers an alternative vantage point to view nuclear deterrence as a risk model rather than strictly as analyzing the probability of a nuclear war event. From this model point of view, risk of nuclear war may be reduced with higher certainty by measures which limit the impact of a nuclear strike rather than relying on inherently uncertain calculations about a rivals intentions.