In 2003, the City of New Haven, Connecticut administered a test to determine who in its Fire Department would be eligible for promotion. The results revealed a striking racial imbalance: although a significant fraction of the applicants were African-American or Hispanic, few if any of these candidates would be promoted. City officials concluded that, in light of this distribution, using the test results would constitute indirect discrimination against African-Americans and Hispanics—leaving them little choice but to invalidate the test. A group of white firefighters who had performed well on the test then sued the City, alleging that by discarding the results, New Haven had engaged in direct racial discrimination against them. In a landmark 2009 decision, the U.S. Supreme Court sided with the white firefighters by a vote of 5-4. Whatever the City’s motives, the Court majority concluded, it had discriminated against the plaintiffs on the basis of race.

This case gives dramatic expression to fundamental conceptual questions about discrimination. When does an effort to avoid indirect discrimination constitute direct discrimination in its own right? Do unyielding prohibitions on direct and indirect discrimination sometimes place agents in a genuine dilemma? And what connection, if any, does an effort to avoid indirect discrimination bear to more familiar efforts to promote racial equality through direct positive discrimination? I offer an account of indirect discrimination that speaks to these questions.