There is a growing body of research examining the labor market returns to college major, motivated by the large returns to skill in the labor market. Prior research has focused almost exclusively on mean effects and has paid little attention to the role of earnings growth and variability. Using linked administrative data from Texas on public K-12 students followed through college into the labor market, we find that the focus on mean differences mask four important features of the returns to college majors. First, majors are associated with varying earnings growth, which makes the returns sensitive to the experience distribution of the sample analyzed. Second, average earnings effects vary across workers; quantile treatment effect estimates show that mean effects mask considerable effect heterogeneity. Third, major choice affects earnings variability within workers over time. College major effects on earnings and variability are negatively correlated; high return majors also have more stable earnings. Finally, there is substantial variation in returns across specific majors within aggregate major groups and across institutions. This variation suggests that estimate of returns to college major are sensitive to how majors are aggregated and the composition of institutions in the sample.