Abstract: Characteristics of the built environment include the arrangement of land uses, transportation infrastructure, and neighborhood design. Built environment policies are hypothesized to influence congestion- and pollution-causing auto use by affecting the convenience of travel by different modes. But after decades of empirical research measuring the strength of built environment influences on travel, theories remain poorly articulated and empirical research contradictory. In the first essay, I argue that the built environment can be expected to influence the quality, quantity and price of traveling to out-of-home activities. Travel choices depend crucially on how the built environment simultaneously affects characteristics of travel by all modes. The proposed theoretical framework provides a means of organizing common hypotheses about the built environment and travel, leading to an increased understanding of interrelationships, better interpretations of existing studies, and improved empirical strategies for future research. The second essay is an empirical study on how development density influences out-of-home non-work activity participation and auto use, using data from a survey that I carried out in San Diego and the San Francisco Bay Area from November 2003 to April 2004. Density's most significant influences are not on the ease of reaching activities on foot or via transit, but on the difficulty of using the auto and, to a lesser extent, on the quality of walking. Previous contradictory empirical findings likely suffer from omitted variable bias. The third essay investigates the hypothesis that densely-developed neighborhoods with walking access to retail shops, good transit service, and finely-meshed street grids attract households who prefer to walk and take transit. If so, observed relationships between the built environment and travel may reflect a residential sorting process rather than an exogenous effect of the built environment on travel. I theorize that residential self-selection is imperfect with respect to non-work accessibility, and show that accounting for pre-existing travel-related preferences increases the explanatory power of statistical models but reveals relatively little bias in conventional analysis.