The major theme of this study is to demonstrate the importance of including the underemployed in assessments of ethnic stratification. Particularly, it is argued that this approach presents a more balanced evaluation of the degree of labor force integration of the Mexican origin population. Empirically, the prevalence of underemployment among Mexican origin and whites workers for the 1976-1987 period is analyzed. During this 12-year span, the level of underemployment for Mexican workers increased from 32 percent to 42 percent, while the white level fluctuated around 21 percent. Models to determine the causes of underemployment were estimated using logistic regression. The models evaluate the relative importance of human capital variables (e.g., age and schooling) and structural variables (e.g., occupation and industrial sector location) as determinants of underemployment. Results indicate that young, poorly educated Mexican origin workers are more vulnerable to underemployment than their white counterparts; that Mexicans incumbents in service and low-skill, blue-collar occupations are more susceptible to underemployment than whites in the same occupations; and that the risk of underemployment is greater for Mexicans than their white counterparts located in the periphery and trade sectors. It was also found that nativity status has no statistical effect on the propensity for underemployment. But Mexicans with limited English proficiency were more likely to be underemployed than their more English proficient peers. To gauge the effect of labor underutilization on ethnic inequality, the Mexican-white earnings differential is analyzed controlling for employment instability. Results show that employment instability exerts a heavy cost: Mexican men who experience employment instability earn 37 percent less than their adequately employed counterparts. Last, the underemployed should be included in ethnic stratification studies focusing on labor market outcomes because their exclusion provides a more conservative assessment of inequality. Data for the analyses come from the Current Population Survey and the 1976 Survey of Income and Education.