A significant challenge to empirically testing theories of discrimination has been the difficulty of identifying taste-based discrimination and of distinguishing it clearly from statistical discrimination. This paper addresses this problem through a two-part empirical test of taste-based discrimination. First, it constructs measures of revealed preferences, which establish that World War I created a strong and persistent shock to ethnic preferences that effectively switched the status of German Americans to an ethnic minority. Second, the paper uses this shock to ethnic preferences to identify the effects of taste-based discrimination at the example of traders at the New York Stock Exchange (NYSE). A new data set of more than 4,000 applications for seats on the NYSE reveals that the War more than doubled the probability that German applicants would be rejected (relative to Anglo-Saxons). The mechanism of taste-based discrimination is surprising: Prices are unaffected by ethnic preferences, and discrimination operates instead entirely through admissions.