We study the impact of the 1930s HOLC residential security maps on experienced segregation based on cell phone records which track visits out of and into home neighborhoods. We compare adjacent neighborhoods, one of which was assigned a lower grade for creditworthiness than the other. We use a sample of neighborhood borders which, based on estimated propensity scores, are likely to have been drawn for idiosyncratic reasons. Neighborhoods on the lower graded side of the border are associated with more visits to other historically lower graded destination neighborhoods. Today, these destination neighborhoods tend to have lower household income and, in some cases, lower educational attainment. We find that these disparities in visits are not driven by work commutes, very local visits, or differences in income. We also find similar disparities for incoming visits. Finally, we study the impact of the maps on non-residential segregation at the city level, based on a comparison of cities around a population cutoff that determined whether a city was included in the HOLC program. Using transition matrices, we describe visit probabilities across the distribution of home and destination neighborhood incomes. In cities with HOLC maps, visits across neighborhood income lines are less common, but this effect is less pronounced for the richest home neighborhoods. These findings suggest that these historical “redlining” maps affect non-residential segregation and the social interactions of urban residents in the present day.