The theories used to understand and predict regular non-problem gambling are almost exclusively affective or cognitive-oriented. These include motives, self-esteem, image enhancement and illusions of control over random events. However, gambling is one of the most frequently purchased consumer products, and the frequency of past behavior has traditionally been viewed as “habit” by psychologists and marketers. While habit as the frequency of past behavior has been shown to be a strong predictor of future behavior in gambling, habit offers little additional insight into gambling behavior in that form.The frequency of past purchasing behavior is an important input to NBD-Dirichlet models that provide an enhanced ability to understand and predict future purchases of frequently purchased consumer package goods. NBD-Dirichlet models have been shown to provide an excellent fit to data for a broad range of frequently purchased goods and services for countries across the world. Applications of the NBD-Dirichlet models to data concerning gambling behavior show that these models consistently provide an even closer fit to the data than with other consumer models tested.The interpretation of NBD-Dirichlet output can provide more accurate benchmarks than cognitive or affective output to test changes to the gambling environment (e.g., more games, new games, warnings) and to gamblers (e.g., problem gambling). The implications and use of the NBD-Dirichlet statistics for gambling providers and public policy is discussed.