Segregation between social groups is a consequential feature of human societies. Increasingly, popular and scholarly accounts of politics in the United States paints a picture of stark partisan segregation, with Democrats and Republicans living separate lives, resulting in partisan rancor and threatening the functions of the democracy. Yet, despite these claims, segregation is bluntly measured, usually relying on aggregate Census data or sparse surveys, often characterized by error so that the levels of exposure across parties is obscured. As such, the important details of partisan exposure have been lost, scholars have only been able to answer whether people tend to share a large geographic area, with members of the other party, not whether that person has members of the other party living in any close proximity, such as a neighbors. We measure partisan bubbles – how far a person must travel in residential context before being exposed to a person of the other party – for every registered voter in the United States. With this dataset of over 180 million geocoded voters, we are able to measure segregation at the individual level across arbitrary levels of geography. At any level of neighbors, we demonstrate how far a Democrat (Republican) must travel to find another Republican (Democrat), or to be in a minority (majority). We then show how this varies by context, measuring differences in partisan segregation across rural and urban areas, and red and blue states.