Empirical studies of markets in disequilibrium have relied on the appropriateness of explicit price adjustment equations, serial independence, normally distributed errors, and explicit equations relating the observed quantity transacted to desired supply and demand. For example, the asymptotic properties of "disequilibrium" estimators and test statistics are sensitive to the parametric forms chosen for price adjustment, the serial behavior of the observations, error distributions, and the quantity transacted. In a word, "disequilibrium" estimators and statistics are non-robust. Unfortunately, economic theory provides little basis for choosing the parametric forms. A lack of economic-theoretic restrictions coupled with non-robust estimators and statistics has severely limited empirical studies of markets in disequilibrium. This dissertation develops new methods for more meaningful estimation of disequilibrium models. The new methods involve more general models and robust estimators. A switching regression model with imperfect sample separation is used to incorporate price adjustment into a disequilibrium model. The model enables price adjustment to be incorporated with less a prior information than usual. To estimate the model, maximum likelihood and least squares estimators are proposed. The asymptotic properties of the maximum likelihood estimator are examined. Previous results for maximum likelihood estimators of disequilibrium models are generalized with asymptotic theory for serially dependent observations. The maximum likelihood estimator is shown to be consistent and asymptotically normal even if the data are characterized by unknown forms of serial dependence. Asymptotic test statistics are also derived. The methodology is illustrated with an empirical application to the U.S. commercial loan market from 1979 to 1984. Finally, I propose semiparametric models and estimators for markets in disequilibrium. These methods are applicable when the error distributions are unknown, and the quantity transacted is an unknown function of supply and demand. Consistent estimators are derived using the method of maximum score.