Racial segregation still prevails in most U.S. cities, but it varies widely across the nation, according to a report from City Observatory.

The murder of George Floyd by police has reignited national interest in making more progress toward racial justice. It’s prompted a new round of introspection about the racism that’s deeply embedded in many American policies and institutions. One pervasive and lingering hallmark is racial residential segregation: our cities have literally been divided by race, and as numerous studies have shown, this has undercut opportunity, perpetuated poverty, limited economic mobility and eroded Black wealth. Today we take a closer look at racial segregation in the nation’s largest cities.

Which cities are the most (and least) segregated?

The most common index of racial segregation is the dissimilarity index, which measures the extent to which different groups of people live in different neighborhoods in a city or metro area. The index ranges from zero (perfectly integrated, where the composition of each neighborhood matches the composition of the larger region) to one (completely segregated) where each neighborhood consists entirely of persons of a single racial or ethnic group). The dissimilarity index expresses the percentage of the population that would need to move to a different neighborhood in order for each neighborhood’s racial/ethnic composition to match that of the larger area.

The Census Bureau’s American Community Survey annually collects data on the race and ethnicity of Americans by Census Tract (a geography that corresponds roughly to neighborhoods). The St. Louis Federal Reserve Bank has used this data to compute the white-non-white dissimilarity index for each of the nation’s counties. Data cover the years 2009 through 2018, and are based on rolling five-year ACS counts (i.e. the 2018 data are drawn from the years 2014–2018). As the Federal Reserve Bank explains:

The Racial Dissimilarity Index measures the percentage of the non-Hispanic white population in a county that would have to change census tracts to equalize the racial distribution between white and non-white population groups across all tracts in the county.

We’ve assembled these data for the central county in each of the nation’s largest metropolitan areas, and then ranked them from least segregated to most segregated. Our tabulation includes only central counties with populations of 100,000 or more. (As we’ve noted, counties are less than perfect units for making these comparisons; metro-area data are more indicative, but the Federal Reserve’s tabulations only address counties).

America’s Least (and Most) Segregated Urban Counties

This chart ranks cities from least segregated to most segregated using the white/non-white dissimilarity index for the largest county in each city’s metropolitan area. The median large metro area has a dissimilarity index of 45, meaning that about 45% of a city’s population would have to move to balance the composition of individual neighborhoods to the region’s overall demographic composition. About half of all large cities have dissimilarity indices between about 38 and 54.

The cities with the highest levels of segregation, according to this measure, are Detroit, New Orleans, Philadelphia, Buffalo and Milwaukee. Each of these cities has a dissimilarity index exceeding 60.

The cities with the lowest levels of segregation are Portland, Oregon, Virginia Beach, Virginia, Boston, Seattle and Las Vegas. Each of these cities has a dissimilarity index of 35 or less. Portland (defined in this case as its largest urban county, Multnomah County) has, by a wide margin, the lowest level of white/non-white segregation of any large urban county in the United States. Its index value is 27.3, about half the median value of the typical large metro in the U.S.

With an index value of 52.01, Houston (defined as Harris County) has the 18th highest level of white/non-white segregation among urban counties in the U.S.

Census data also enable us to plot the pattern of segregation over time. Small year-over-year variations most likely reflect sampling variability from the American Community Survey, so it’s best to look at multi-year trends. For most large U.S. metro areas, the trend in segregation is downward: Dissimilarity indices are declining over time. Here’s a chart showing Houston (Harris County) white/non-white dissimilarity from 2009 through 2018:

Over this decade, Portland’s white/non-white segregation index declined from 31.6 to 27.3. Segregation in Portland has been declining recently, but it’s actually a trend that’s been in place for many decades. During that period, Houston’s white/non-white segregation index declined from 52.5 to 52.01.

Portland’s big decline in segregation

Portland has not always been a highly integrated place. If we look at the historical data on the Black-white segregation index for Portland for the period 1970 through 2010, we see that Portland went from being one of the most racially segregated metro areas to one of the least. (We use the Black-white measure because the Census Bureau’s race and ethnic definitions were not comparable in early years. Data compiled by Sophie Litschwartz at the Urban Institute shows that in 1970, Portland was more segregated than the typical large metro area, but that segregation has declined sharply since then. Portland’s Black-white segregation measure fell by half (40 points) in 40 years; while the median rate for large metro areas fell by about 15 points.

It seems odd that Portland would be a leader in integration, given that it is by many media accounts and popular wisdom “the whitest city in America.” While it’s true that the city has relatively few Black residents compared to most large American cities, as many demographers point out, that particular definition of “white” effectively treats Hispanic, Asian, Native American and mixed-race people as white. When you look at the share of the population that is “Non-Hispanic white,” Metro Portland is more diverse than Cincinnati or Pittsburgh, for example.

A city’s racial and ethnic mix is a product of history and geography. The geography of Latinos, Blacks and Asians in the United States each have their own geographic contours based on historical patterns of migration. Blacks were brought to the Southern United States as slaves, landed primarily at places like Charleston, Tidewater Virginia and the Gulf Coast, and to this day, are disproportionately concentrated in the South. Most Latinos have migrated from Latin America, and are heavily concentrated in the Southwest.  Asians have a disproportionate concentration on the U.S. West Coast. The patterns that were in place 100 years ago are still reflected today in the regional concentrations by race and ethnicity.

What’s more malleable to change in the short run (over the course of a few decades) is where within metropolitan areas people live. In general, throughout the United States, we’ve seen a marked decline in residential racial segregation. Localized patterns of segregation can change more quickly than the overall racial and ethnic diversity of a metro area.  Given its more limited overall racial/ethnic diversity, Portland achieves a higher level of integration than nearly all US metro areas, something we explored in our report “America’s Most Diverse, Mixed Income Neighborhoods.”

Greater diversity is already baked in Oregon’s demographic cake. Demographer Charles Rynerson points out that–

“2019 Census estimates and found that people of color make up just 10% of Oregonians 65 or older. But they are 37% of those under the age of 15.”

This means that the state will become progressively more diverse with each passing year.

This post originally appeared on the City Observatory website. Joe Cortright is the director of City Observatory.