To explore the limits of quantification as a form of rationalization, we examine a rare case of dequantification: race-based affirmative action in undergraduate admissions at the University of Michigan. Michigan adopted a policy of holistically reviewing undergraduate applications in 2003, after the U.S. Supreme Court ruled unconstitutional its points-based admissions policy. Using archival and ethnographic data, we trace the adoption, evolution, and undoing of Michigan's quantified system of admissions decision-making between 1964 and 2004. In a context in which opponents of the system had legal avenues to engage a powerful outside authority, we argue that three internal features of the University's quantified admissions policy contributed to its demise: its transparency, the instability of the categories it quantified, and the existence of qualitative alternatives. Our analysis challenges the presumed durability and inevitability of quantification by identifying its vulnerabilities and suggests that quantification should be understood as a matter of degree rather than a simple binary.