The internet provides an unprecedented variety of opportunities to people. Whether looking for a place to go on vacation, an apartment to rent, or a PC to buy, the potential customer is faced with countless possibilities. Most people have difficulty finding exactly what they are looking for, and the current tools available for searching for desired items are widely considered inadequate. Search engines can be very effective in locating items if users provide the correct queries. However, most users do not know how to map their preferences to a query that will find the item that most closely matches their requirements. In this thesis, we aim at supporting users to make sound decisions while accessing an online electronic catalog, a task that we call preference-based search. We consider how biases typical of human decision making can arise with traditional web tools (as forms that ask the user to answer a list of questions). With user studies we show that common interfaces induce the users to state incorrect preferences due to means-objectives, leading to poor decision accuracy. In many cases, users searching for products or information are not very familiar with the available items and their characteristics. According to behavioral decision theory, their preferences are not well established, but constructed while learning about the possibilities. We explore the use of interactive search tools for helping the user make accurate decisions. In particular, we consider at example-critiquing, a technique for enabling users to incrementally construct preference models by critiquing example options that are presented to them. We present novel techniques for improving the example-critiquing technology by adding suggestions to its displayed options. Suggestions have the goal of making the user aware of its true preferences, supporting the psychological process of preference construction. According to our experience, most of the preferences (79%) emerge from positive critiques that identify an opportunity that the user had not considered before. Our model-based suggestions are produced based on an analysis of users' current preference model and their potential hidden preferences. The intuition behind this approach is that suggestions should be options that have a high probability of becoming optimal when a new preference is stated. The uncertainty over the user model is represented by probabilistic distributions over the possible preferences that the user might have. We evaluate the performance of our model-based suggestion techniques with both simulated and real users. User studies showed that interactive tools with suggestions provided by my model achieve higher decision accuracy. We consider how to improve the interaction with the user, where at each cycle of the interaction the system provides suggestions that are adapted, learning from the user's past actions and can consider prior knowledge. We discuss how to efficiently implement preference-based search in practice, both in databases and in configurable catalogs (where the options are constructed from a set of constraints; the set of feasible configurations can be extremely large).