This paper uses NLSY data from 1979-1993 to estimate training's effect on one year wage growth. Year-by-year training histories are constructed which allow the returns to training received at both current and previous employers to vary over time. The time patterns of the returns to training are constructed for both long and short spells of training over nine and three year periods respectively. These returns are then estimated for different demographic groups in order to see how education level, test scores, and occupation influence the payoff to training. Both company training and formal schooling were associated with significant wage growth even nine years after they occurred. Company training was associated with significant wage growth effects irrespective of whether workers changed jobs, although wage growth was higher when the training occurred at a previous employer. Contrary to the conventional human capital model, employers appear to be sharing the costs and returns of general training. While training incidence was lowest for high school dropouts, their return to getting training was the highest. College graduates, in contrast, received the most training but benefited the least. These results suggest an under-supply of training opportunities for low skilled workers.