In order to evaluate demands for new forms of residential environments (such as cluster development rather than the spread pattern of suburban sprawl), it may be useful to deal with persons' residential preferences directly, rather than their market choice. The paper develops residential environmental preference variables from questionnaire survey data. The preferences may be considered as relative tradeoffs among the residence's accessibility, characteristics of the house and lot (for example, lot size), the quality of the natural environment, and characteristics of the non-natural environment (for example, population density). At the same time, comparable measures are devised of the characteristics of the persons' present residences—their residential choice. Principal results are that the preference and choice variables do seem useful; that preferences, thus defined, do express themselves to some extent through market choice; and that different persons do in fact prefer different residential characteristics. Preferences for different residential characteristics tend to be negatively associated, so that there are persons with strong preferences for the natural environment who are willing to sacrifice a large lot to live in a beautiful area.