This dissertation studies consumer search behavior in markets where buyers have incomplete information about available goods, such as markets with many sellers or frequently changing prices. In these markets, consumers engage in costly search in order to collect information necessary for making a purchase. Our method of investigation combines real-world data on consumer search with explicitly formulated economic models that predict the behavior of a rational consumer in such situations. Using this method, we focus on two sets of questions. First, we are interested in the identification of the model of rational search, i.e. whether or not its parameters can be uniquely recovered using the available data. Second, provided that the model is identified, we estimate the model to study the implications of the search process for consumer demand. In Chapter 1, we provide an introduction to the subject and review some of the relevant existing literature. In Chapter 2, we estimate a model of search where consumers are looking for hotels online. For that purpose, we use a unique data set on individual search histories by consumers who visited a search website. We show that the model is non-parametrically identified, given our data. On the same data set, we also estimate a static model that ignores the search process and explains only the final purchase. By comparing demand estimates between the two models, we find that the static model over-estimates the price elasticity of demand by four times. This means that search frictions have significant implications for consumer demand and should be accounted for in estimation. The median search cost is estimated to be 38 dollars per page of 15 hotels. In other words, the median consumer is indifferent between getting 38 dollar discount now or studying another 15 hotel options, in a hope to find a better deal later. In Chapter 3, we continue working with the same data set on hotel searches. We enrich the model of the previous chapter in the following way. While searching, consumers not only learn about new accommodation offers, but also infer the price-quality relationship that currently exists on the market. That is, the consumer is initially uncertain about the premium she has to pay for an additional star rating, and learns about this premium in a Bayesian fashion as new price quotes arrive. In this way, we relax an assumption commonly used in the existing empirical applications of consumer search: that consumers know the price distribution and therefore do not take into account the information collected during the search process. In addition to being more realistic, the learning mechanism allows us to make inferences about consumer prior beliefs. The identification comes through the joint variation of information sets and search actions across consumers in the data set. We estimate models with and without learning on the data set of hotel searches and perform a statistical test between the two approaches. We find that the data favors the learning hypothesis. We also find evidence that consumers underestimate the price of quality, relative to the relationship found in the actual data. In Chapter 4, we remain in the framework of search where consumers learn about the price distribution while searching. For a particular class of prior beliefs, called the Dirichlet distribution, we develop a novel characterization of the optimal stopping rule. An advantage of our representation is that it delivers closed form, easily computable formulas for ex-ante purchase probabilities, conditional on consumer preferences and search costs. Such formulas are necessary for incorporating search frictions into the estimation of demand in cases where only aggregate purchase data is available. Indeed, the kind of search data we used in the previous chapters is still rare. We apply our method on a dataset of prices and market shares of S&P 500 mutual funds. From the model's estimates, we find that the top funds have lower market shares and lower price elasticities if consumers are learning about the price distribution than if they do not. In Chapter 5, we conclude and provide directions for future research. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]