In the literature on search markets, agents know the distribution of prices or learn about it through short run Bayesian updating, even though the first assumption is unrealistic and the empirical evidence emphasizes learning in the intermediate and long runs. In this paper, a new search markets theory is developed where agents have beliefs about the distribution of prices based on finitely many of its moments and past market experiences. The main results are the existence of a Hicks–Grandmont temporary equilibrium, that the equilibrium correspondence has a closed graph, and sufficient conditions for price dispersion. A simple example illustrates the main ideas. Journal of Economic Literature Classification Numbers: D83, L13.