This paper uses repeated cross-section data from Norway to estimate the demand for fruits and vegetables (FV) and physical activity (PA) with a particular focus on the role of socioeconomic status. Conventional econometric count data models produce results that are commonly found in empirical work; the effect of higher socioeconomic status on healthy behavior is positive and generally statistically significant, but the average partial effects are in some cases small and imprecisely estimated. For both behaviors, subsequent latent class models identify two subpopulations - or groups of people - with different sets of preferences; one group has low latent demands, but for these individuals, average partial effects of socioeconomic status are generally stronger than those predicted by the conventional models. The other smaller group consists of individuals who have high latent demands, but whose variability in behavior is poorly explained by socioeconomic status. Posterior analysis shows that individuals with higher socioeconomic status are more likely to belong to the healthier of these two groups. Proxies for time preferences, risk, self-control, and time constraints are also found to be important in characterizing high latent demand groups for PA and FV.