Abstract: The emerging multiracial population charges social scientists to reevaluate prevailing theories of race, which has largely been considered singular and fixed over the span of one’s life. In this study, I explore the relationship between racial context and partisanship for Black-White biracials. In doing so, I also investigate the ways that racial group membership may moderate the influence of context on political outcomes. I argue that racial group boundaries are more permeable for biracials than for monoracials. As a consequence, biracials should have more latitude than monoracials in their ability to adopt the partisanship of different racial groups as they vary contextually in salience. To test these claims, I apply a unique matching algorithm to identify Black-White biracial voters in the North Carolina voter file. The unique construction of this dataset largely elides issues of residential self-selection, and I employ double robust estimation with machine learning to derive potential effect estimates. I find that Black-White biracials vary so extremely in their partisanship across racial context that they are statistically non-different from monoracial Blacks when in heavily Black social contexts, and non-different from Whites in White contexts. I replicate these findings using four nationally-representative survey datasets and a host of variation in methods and measurement. Overall, this study helps clarify the linkages between race, racial identity, and politics, and substantiates that racial context plays a critical role in biracials’ politics.