High school grade point average (GPA) is perhaps the single most important determinant of student access to post-secondary opportunities: several states employ policies that guarantee admission to a public university based on class ranking; admissions offices use GPAs to rank applicants; and many colleges and scholarship organizations have minimum GPA requirements.1 Such policies give equal weight to performance at each high school grade level, implicitly assuming that each grade is equally important for predicting future success. In contrast to this assumption, this paper presents evidence that a student’s performance in later grades has far more power for predicting college dropout, on-time graduation, and future earnings than does performance in prior years. Evidence is presented that the predictive power of later high school grades stems in part from persistent trends in student effort. These findings suggest that using grade level GPAs may result in the allocation of enrollment slots and financial resources to students who have higher probabilities of success. Likewise, extending application deadlines would allow the consideration of 12th grade performance, which is most predictive of long-run outcomes. The results have important policy implications in light of the labor market returns associated with college completion and the high cost of increasing enrollment and persistence through financial aid and grants.2 Data on high school course grades and state university applications for four Florida cohorts reveal that admissions during the period of study were based on grade point averages, with nearly identical weight given to performance in 9th, 10th, and 11th grades (and less to 12th grade performance, which is only partially revealed at the time decisions are made). This balance across grade levels is consistent with the common practice among colleges, and large universities in particular, of constructing an admission index using cumulative high school GPA and entrance exam score.3 However, giving equal weight to each grade level necessarily discards a large amount of information stemming from substantial variation in student performance over the four years of high school. For example, within-student changes in GPA between 9th and 12th grades have a standard deviation of 0.7 points. Among students in the middle quintile of performance in 9th grade, 11% are in the top quintile by 12th grade while 15% are in the bottom quintile.