Models in which employers learn about the productivity of young workers, such as Altonji and Pierret (2001), have two principal implications: First, the distribution of wages becomes more dispersed as a cohort of workers gains experience; second, the coefficient on a variable that employers initially do not observe, such as the Armed Forces Qualification Test (AFQT) score, grows with experience. If employers' learning varies significantly across occupations, both of these indicators of learning should covary positively across groups defined by a worker's occupational assignment at labor market entry. This paper tests this implication of the employer learning model using data from the NLSY and CPS. I find that occupations with high growth in the variance of residual wages over the first ten years of the worker's career are also the occupations with high growth in the AFQT coefficient, confirming the learning perspective. Interestingly, occupations that my analysis characterizes as having a low level of employer learning are not occupations where employers know little about the worker after ten years of experience; instead they appear to be occupations where employers have already learned about the worker's AFQT score at the time of hire. I provide several pieces of evidence that occupational assignment affects the learning process independently from education and that the results are not driven by workers' occupational mobility.