In this paper, we extend existing models that use the NLSY 79 to document employer screening and learning by showing that the return to education and ability change with experience. Specifically, we test for and document a non-linear relationship between wages and ability as measured by the AFQT score at low levels of potential experience. For high levels of AFQT, wages appear to fall as AFQT increases. As experience increases, the relationship between wages and AFQT returns to a monotonic relationship. As a result much of the observed increase in the return to AFQT as potential experience increases is associated with a change in the shape of the relationship, and the increase in the return to AFQT at lower levels of AFQT is more modest. These results are robust using samples and models from previous papers on the subject, developing a broader sample using all waves of the NLSY 79, and analyzing the question using data from the NLSY 97. Finally, we find evidence that high AFQT workers without four years of college select into occupations that provide more training, perhaps sacrificing initial wages in order to build skills.