The potential use of actual performance by high school students in college-level courses as a predictor of college achievement, rather than or in addition to Scholastic Aptitude Test (SAT) scores, was investigated. At Syracuse University Project Advance over 4,000 students a year in 80 high schools in four states participate in joint high school and college during their senior year in high school. Grade point average (GPA) in Project Advance courses was a a predictor of first semester college GPA in a stepwise multiple regression analysis for 1978, 1979, and 1980 Project Advance students who subsequently attended Syracuse University. Percentile of high school rank, SAT math scores, and SAT verbal scores were also analyzed. SAT verbal vocabulary scores and SAT verbal reading scores were included in a separate analysis for those students who had the additional SAT scores available. Consonant with past research, percentile of high school rank was a good predictor and SAT scores were not. The correlation between Project Advance GPA and first semester GPA was moderately high. Project Advance GPA was the best single predictor in one year, added significantly to percentile of high school rank in another year, and was not helpful at all in the third year. Overall, there was a decline in the amount of variability accounted for over the three-year period. It is concluded that high school performance is the best predictor of college achievement, and SAT scores are not very useful. The amount of variability accounted for by different predictors varies widely from college to college; therefore, colleges may want to determine the best predictor equation for admission criteria. (Author/SW)