A number of states have recently prohibited the use of affi rmative action in admissions to public universities statewide. A growing body of research suggests that these affi rmative action bans reduce minority enrollment at selective colleges while leaving overall minority college enrollment rates unchanged. The effect of these bans on racial segregation across colleges has not yet been estimated directly and is theoretically ambiguous due to a U-shaped relationship between minority enrollment and college selectivity. This paper uses variation in the timing of affi rmative action bans across states to estimate their effects on racial segregation, as measured by exposure and dissimilarity indexes. The results suggest that affi rmative action bans have in some cases increased segregation across colleges but in others cases may have actually reduced it. In particular, early affi rmative action bans in states with highly selective public universities appear to be associated with more segregation, whereas other affi rmative action bans appear to be associated with less segregation.