On the Religious Population

From 1850 to 1946, the U.S. Census counted religious communities and structures. If one is aware of the history of the 20th Century, and the role played by Hollerith’s machines in that history, what other option was there than to stop counting after the war? Besides, it might be impolite.

There are many reasons people look at these numbers, but one is “Why is my church shrinking?” This is not the same question as “Why is my church declining?”

Generally speaking, the answer to the former is twofold, and does not involve theological arguments: transportation and demographics. People have cars. People left your town. And your congregation isn’t having children. (As just one example, in 1970, the Sunday School at Grace Episcopal Church, Millbrook, NY, had 100 children enrolled; in 2023, it has 4, more or less.) 

The answer to the latter includes the former, but also the distribution of wealth. In particular cases it may include theology and practice.

The Census data conveniently allows us to expand our statistical view of religious practice beyond the last few decades given in most graphs. And so here are local organizations and edifices as a rate of the population.

Americans rapidly formed new religious communities and built physical homes for them between 1870 and 1900 — the number of congregations and buildings more than doubled between 1870 and 1890—yet by 1930 the proportion of each relative to the population was roughly the same as in 1850. Notably, the above does not show individual persons, but buildings and congregations, so we cannot estimate the size of these congregations. Many of the structures remain and architectural historians could provide an estimate of seating capacity. However, the Census also reported membership from 1890 to 1936, and compiled membership numbers provided by other sources thereafter, so we’ll just use those. In 1890, 34% of the population was a member of a religious body. By 1920, 51% were. This dropped slightly during the roaring Twenties, but by 1960 was up to 64%. By 2020, this was 49% — which is low only by comparison with 1960.

Assuming the accuracy of the data. The Census notes that “data presented are not directly comparable from census period to census period.” Where they overlap, these numbers contrast sharply with those from the Yearbook. The ARDA numbers don’t overlap. And there’s a large 15% gap between 1970 Census numbers and 1980 ARDA numbers, which probably reflects the different survey methods.

So, how many people per congregation is that? On average, 131 members per congregation in 1890, to 452 per congregation today, down from a peak of 622 in 1970. But that’s an average. A dive into the statistics of a given body would show Pareto’s long-tailed curve: many tiny churches, a few enormous. However, we can safely surmise that there are fewer, larger churches now than in the past. Why? Because the number of congregations as a rate of the population is lower, yet the average size is higher.



  • 11th Census of the United States 1890, Volume IX: Report on the Statistics of Churches in the United States https://www2.census.gov/library/publications/decennial/1890/volume-9/1890a_v9-01.pdf
  • Historical Statistics of the United States: Colonial Times to 1970, Part 2, Chapter H: Social Statistics, Series H 788-792, H 793-799 https://www2.census.gov/library/publications/1975/compendia/hist_stats_colonial-1970/hist_stats_colonial-1970p1-chH.pdf
  • Association of Statisticians of American Religious Bodies (ASARB), The Yearbook of American & Canadian Churches (The Yearbook), https://www.yearbookofchurches.org/interactive-charts-maps. The Yearbook was formerly published by the National Council of Churches of Christ in the United States, who began data collection in 1916.
  • The Association of Religion Data Archives, U.S. Religion Census, https://thearda.com/us-religion/census/congregational-membership and https://www.usreligioncensus.org