This research examines long-run changes in alcohol consumption patterns for the United States, at the state level, and presents forecasts for per capita consumption of beer, wine, and spirits developed using the ARIMA methodology. The evidence is then presented on the extent of convergence in consumption through time. This evidence shows that from the 1970s through the early 2000s, a pattern of convergence in both the level of consumption and the consumption mix was evident, but since the early 2000s, and unlike the pattern observed globally, there has been a reversal of this trend. The changes in consumption through time are illustrated via ternary plots. Bayesian estimation methods are used to for- mally describe changes in historical consumption patterns and to investigate the impact of policy settings on consumption forecasts. There were no systematic correlations found between alcohol policy settings and forecast future consumption changes, or tax rate levels and forecast consumption changes.