Identifying the attributes that households consider when making residential location decisions is central to land use and transportation planning and policy analysis. That different households have different tastes for these attributes is axiomatic; this research focuses on identifying and representing the differences. In discrete choice applications, including the analysis of residential location decisions and travel mode choice, taste variation can be addressed by segmenting the sample using indicators of household structure (age of household head, household size, and presence of children) and household sociodemographic variables (income, race, and gender) and by developing model specifications that allow taste weights for attributes to be influenced by household structure and sociodemographic variables. This research draws on multi-disciplinary residential location literature to identify potential sources of taste variation and employs various analytical techniques, including full and partial segmentation, to represent systematic variations in taste for housing, population, and accessibility attributes among households in the San Francisco, California, metropolitan area. This research addresses the two major deficiencies of most residential location choice model specifications, namely (1) the omission of important tract socioeconomic status, family suitability, and racial composition attributes from the utility function and (2) the omission of taste parameters that identify differences in taste for attributes for households with different characteristics.