When properly understood, scores on the Scholastic Aptitude Test (SAT) support the hypothesis that the performance of American students as measured by the SAT is steadily improving. As the overall pool of SAT takers increases and as pressures to adopt SAT scores as an administrative measure of instructional performance mount, SAT averages at the national, state, and local levels can be expected to be driven down. Linear regression makes weeding out the impact of sample size changes on SAT averages possible. Applying these techniques to 1988 data for the 1.1 million students who took the SAT nationwide, the 913,000 who took the SAT in states where it is the principal college entrance examination, and the 187,000 who took the SAT from other states, reveals real increases in average SAT scores. From 1987 to 1988, there was a 9.1 point increase. Only one state had a decrease in real mathematics performance, and only two had decreases in real verbal performance. Many have accepted the notion of a decline in SAT scores; such a decline can be useful in competition for budgets and recognition. The apparent improvement needs to be clarified; it may be due to students being inherently better, or it may reflect that students are better prepared for taking the SAT. (SLD)