The model includes independent variables and definitions that specify a simple model of entrepreneurship. "Entrepreneurship" is defined as a dichotomous household-level variable that assumes a value of "1" if any member of the household is nonprofessionally self-employed and "0" otherwise. Using logistic regression, "entrepreneurship" is expressed as a function of selected personal, household, and structural characteristics. The personal characteristics pertain to the household head. Characteristics are divided into three basic classes: human resources (age, sex, marital status); human capital (education, foreign higher education, and English ability); and immigration resources (birthplace, citizenship, and year of arrival). Household characteristics are also in three basic categories: human resources (relative numbers of dependents and workers as well as hours of unpaid family labor); human capital (the mean education of household members and their overall English language ability); and financial capital (wage and salary income, interest and dividend income, and home ownership). Structural characteristics are measured at the metropolitan level and include the size of the racial or ethnic group, its poverty rate, and the degree of residential segregation it experiences. Using data from the 1990 U.S. Census, this study used the model to measure the effect of segregation on the likelihood of entrepreneurship among different racial/ethnic groups in U.S. metropolitan areas. The study found that beyond very moderate levels, segregation lowered the odds of entrepreneurship. These findings suggest that black entrepreneurship will be enhanced, not only by programs to improve the socioeconomic circumstances of black households and efforts to lower discriminatory barriers within capital markets, but also by reducing discrimination within housing markets to promote a broader desegregation of American society. 5 tables, 25 references, and appended data on mean characteristics of Hispanic and Asian subgroups