This paper uses newly available Chinese micro data to estimate the return to college education for late 20th century China when allowing for heterogeneous returns among individuals selecting into schooling based on these differences. We use recently developed semiparametric methods to identify the parameters of interest. We demonstrate that heterogeneity among people in returns to schooling is substantial. People sort into schooling on the basis of the principle of comparative advantage, which we document to be an empirically important phenomenon in modern Chinese labor markets. Standard least squares or instrumental variable methods do not properly account for this sorting. Using new methods that do, we estimate the effect on earnings of sending a randomly selected person to college is a 43% increase in lifetime earnings (nearly 11% annually) in 2000 for young people in urban areas of six provinces of China. The effect of college on those who go is 13%. Our evidence, and simple least squares evidence, suggests that after 20-plus years of economic reform with market orientation, the return to education has increased substantially in China, compared to the returns measured in the 1980?s and the early 1990?s.