BSTRACT Fears of rogue states, withdrawal of Cold War-era security guarantees, a falling tech- nological threshold, and concerns about availability to terrorist organizations ensure that nuclear weapons proliferation remains a central security issue and that developing an ade- quate theory of proliferation ranks high on the agenda of international relations scholars. Yet while scholars have offered an abundance of explanations for proliferation decisions, little consensus exists on the adequacy of various theories or whether we even possess a theory of nuclear proliferation. I argue that this unsatisfactory state of affairs derives at least partly from a mis-match between theoretical arguments, which tend to make probabilistic claims and envision multiple causal variables, and empirical methodology, which tends towards case studies implicitly applying deterministic standards with a univariate logic of inference and frequently samples on the dependent variable. Seeking to complement existing research, we construct a new data-set on nuclear proliferation, identifying three different stages on the path to weaponization of nuclear weapons technology, and use partial likelihood hazard models to test theories of nuclear proliferation. The data analysis suggests that existing theories deserve more credit than they are frequently given: nuclear weapons proliferation is reasonably well explained by level of economic development, external threat environment, level of integration in the world economy, and satisfaction with the policies of the leading regional power.