This paper estimates the sensitivity of students’ college application decisions to a small change in the cost of sending standardized test scores to colleges. Using confidential ACT micro data, I find that when the ACT increased from three to four the number of free score reports that ACT-takers could send, the fraction of test-takers sending four reports rose substantially while the fraction sending three fell by an offsetting amount. Students simultaneously sent their scores to a wider range of colleges. Using micro data from the American Freshman Survey, two identification strategies show that ACT-takers sent more college applications and low-income ACT-takers attended more selective colleges after the cost change. The first strategy compares ACT-takers before and after the cost change, controlling for time trends and covariates, and the second estimates difference-in-difference regressions using SAT-takers as a control group. Back-of-the-envelope calculations suggest that by inducing low-income students to attend more selective colleges, the policy change significantly increased their expected earnings. Because the cost of sending an additional (non-free) ACT score was merely $6 throughout, this sizable behavioral change is surprising and suggests that students may use simple heuristics in making their application decisions. In such a setting, small policy perturbations can have large effects on welfare.