Over time have less people lived in American's largest cities?

Claude has a little prospective below.

A few things worth noting in this arc:

The 1930 peak of 19.4% is the high-water mark of the pre-suburban American city. In that year, one out of every five Americans lived in just twenty cities, with New York alone housing 5.6% of the country. The subsequent decline is one of the most dramatic demographic reorganizations in US history — losing nearly half of that concentration over 60 years.

The steepest drop was 1950 → 1980, when the top-20 share fell from 18.2% to 12.1%. This is the era of the Interstate Highway System, mass automobile ownership, the GI Bill, and the Rust Belt collapse. Detroit lost a third of its population, St. Louis lost more than half, Cleveland and Pittsburgh got hollowed out. Even where metros kept growing, the growth happened outside city limits.

Since 1990 the line has essentially flattened at ~10-11%, but the composition has changed radically. In 1930, seven of the top 20 were in the Northeast/Midwest industrial belt. In 2024, only three of the top 20 are north of the Ohio River (NYC, Chicago, Philadelphia). Five are in Texas alone. The category "20 largest American cities" holds roughly the same share of the country, but it's essentially a different set of cities.

One methodological caveat worth flagging: this chart uses city-proper (incorporated place) populations at each census. Sun Belt cities benefited enormously from annexation — Houston went from 44,633 people over ~9 square miles in 1900 to 2.4M over ~665 square miles today, while Boston is still 48 square miles. So the post-1980 flatness partly reflects Sun Belt cities annexing their suburbs, not just growing organically. A metro-area version of this chart would show a very different (and steeper-rising) curve — the top 20 metros probably hold around half the US population today.