This article analyzes the competitive dynamics in global arms transfers from 1951 to 1995. I discuss the enduring forces behind the proliferation of military capability during the Cold War and other historical periods, and then consider the competitive dynamics characteristic of the superpower rivalry itself. The process of military‐technological advance, along with the dynamics of enduring interstate rivalry, lead us to expect certain patterns in quantitative data representing arms‐transfer levels over time. Concepts in time‐series analysis—cointegration and error correction—are helpful for understanding competitive arms‐transfer policies during the Cold War, and I apply the relevant analytical tools to test for the hypothesized patterns in the empirical data. American and Russian, as well as NATO and Warsaw Pact, arms transfers are examined at three levels of regional aggregation: the Third World as a whole, the Middle East security complex, and the Persian Gulf subcomplex. The evidence shows that arms transfers by the Cold War rivals moved together in patterns consistent with competitive policy making in an environment of military‐technological change, and that one or both sides adjusted their supply policies to correct for deviations from a moving equilibrium. This describes an action‐reaction process, but a loosely coupled one deriving from military‐technological uncertainties, the complexities of regional security dynamics, and the multidimensional character of the Cold War competition.