This paper examines the significant changes in the U.S. wage structure over the past fifteen years, focusing on the divergent trends in upper-tail (90/50) and lower-tail (50/10) wage inequality. The authors document that wage inequality in the top half of the distribution has shown a secular rise since 1980, while wage inequality in the bottom half has ceased growing and even narrowed in some measures since the late 1980s. They also find that occupational employment growth shifted from monotonically increasing in wages (education) in the 1980s to a pattern of more rapid growth in jobs at the top and bottom relative to the middle of the wage (education) distribution in the 1990s. This pattern is characterized as the "polarization" of the U.S. labor market, with employment polarizing into high-wage and low-wage jobs at the expense of middle-wage work. The authors propose a model of computerization to explain these observed patterns, suggesting that computers complement non-routine cognitive tasks in high-wage jobs and substitute for routine tasks in many traditional middle-wage jobs, while having little direct impact on non-routine manual tasks in low-wage jobs.This paper examines the significant changes in the U.S. wage structure over the past fifteen years, focusing on the divergent trends in upper-tail (90/50) and lower-tail (50/10) wage inequality. The authors document that wage inequality in the top half of the distribution has shown a secular rise since 1980, while wage inequality in the bottom half has ceased growing and even narrowed in some measures since the late 1980s. They also find that occupational employment growth shifted from monotonically increasing in wages (education) in the 1980s to a pattern of more rapid growth in jobs at the top and bottom relative to the middle of the wage (education) distribution in the 1990s. This pattern is characterized as the "polarization" of the U.S. labor market, with employment polarizing into high-wage and low-wage jobs at the expense of middle-wage work. The authors propose a model of computerization to explain these observed patterns, suggesting that computers complement non-routine cognitive tasks in high-wage jobs and substitute for routine tasks in many traditional middle-wage jobs, while having little direct impact on non-routine manual tasks in low-wage jobs.