By exploiting establishment-level data for U.S. manufacturing, this paper sheds new light on the source of the changes in the structure of production, wages, and employment that have occurred over the last several decades. Based on recent theoretical work by Caselli (1999) and Kremer and Maskin (1996), we focus on empirically investigating the following two hypotheses. The first hypothesis is that the channel through which skill biased technical change works through the economy is via changes in the dispersion in wages and productivity across establishments. The second is that the increased dispersion in wages and productivity across establishments is linked to differential rates of technological adoption across establishments. We find empirical support for these hypotheses. Our main findings are that (1) the between plant component of wage dispersion is an important and growing part of total wage dispersion, (2) much of the between plant increase in dispersion is within industries, (3) the between plant measures of wage and productivity dispersion have indreased substantially over the last few decades, (4) industries with large changes in between plant wage dispersion also exhibit large changes n between plant productivity dispersion, (5) a substantial fraction of the rising dispersion in wages and productivity is accounted for by increasing wage and productivity differentials across high and low computer investment per worker plants and high and low capital intensity plants, and (6) Changes in dispersion accounted for by such observable characteristics yield predicted industry level changes in wage and productivity dispersion that are highly correlated.