This study estimates the earnings differential between college and high school graduates, denoted as the college earnings premium, from 1940 to 1988. The average measured premium exhibits a decline in the 1940s, gradual increases in the 1950s and 1960s, a decline in the 1970s and a rise in the 1980s for younger male workers and most female workers. Overall the results indicate that this differential has remained relatively high during this period, even given the concurrent increase in the supply of college graduates. As a result, estimates in the expected trend in the college earnings premium based on relatively short time periods are likely to be misleading. Although the data is not well suited to explaining the observed fluctuations in the college earnings premium, some support is given to the hypotheses that cohort size and the business cycle can influence it.