This paper is concerned with why immigrants appear to have consistently lower partial effects of schooling on earnings than the native born, both across destinations and in different time periods within countries. It uses the Over-Under-Required education approach to occupations, a new decomposition technique developed especially for this approach, and data from the 2000 Census of the United States. Based on the average (mean or mode) level of schooling in their occupation, the schooling of the native and foreign born adult men is divided into the "required" (average) level, and years of under- or overeducation. Immigrants have a wider variance in schooling, with an especially large proportion undereducated given the average schooling level in their occupation. Immigrants are shown to receive approximately the same rate of return to the "required" (occupational norm) level of education, but experience a smaller negative effect of years of undereducation, and to a lesser extent a small positive effect of overeducation. About two-thirds of the smaller effect of schooling on earnings for immigrants is due to their different payoffs to undereducation and overeducation. The remainder is largely due to their different distribution of years of schooling. The country-of-origin differences in the returns to under- and overeducation are consistent with country differences in the international transferability of skills to the US and the favorable selectivity of economic migrants, especially those from countries other than the English-speaking developed countries.