How might the global share of people who believe in God change by 2100? The Forecasting Machine's 29 May 2026 assessment made a significant decrease the most likely outcome, but only narrowly. The full report assigned 50% to a significant decrease, 31% to no significant change, and 19% to a significant increase.
This is a 74-year forecast about a concept that is harder to measure than population, income, or temperature. It combines long-run evidence on secularization with demographic forces that may sustain religious belief. The result is not a claim that religion will disappear. It is a probability distribution over the direction of the global share.
The forecast at publication
The source cover preserves an earlier model snapshot from 23 May: 45.5% for a significant decrease, 27.5% for no change, and 27% for a significant increase. The later full report moved further toward decline. Showing both timestamps is useful because it makes the evolution of a long-range forecast visible rather than presenting one percentage as timeless.
Why a significant decrease leads
The decline case rests on mechanisms observed across many societies: rising education, urbanization, economic development, individualization, and generational replacement can weaken the social expectation that religious belief is the default. Over several generations, even gradual change has time to compound.
The forecast also considers diffusion. Regions that remain highly religious today are not insulated from the forces that changed belief elsewhere. Expanding access to education and digital information, changing family structures, and younger cohorts with different attitudes can gradually alter high-belief populations as well as already secular ones.
Why the world may remain more religious than expected
Demography pushes in the other direction. Populations in parts of Sub-Saharan Africa and South Asia are growing more quickly and often report high levels of religious identification and belief. Meanwhile, many highly secular societies have low fertility. Those differences can slow or offset secularization in the global average even if belief declines within individual countries.
Pew Research Center's demographic projections make this counterweight concrete. Their work projects religious affiliation using age, fertility, mortality, migration, and religious switching, and finds that the world's religious composition can move differently from trends in Western Europe or the United States. That evidence does not directly answer this forecast, because affiliation and belief in God are not the same variable, but it shows why demographic structure cannot be ignored.
- Secularization: modernization and generational replacement can reduce belief over time.
- Demographic composition: faster growth in high-belief regions can lift their weight in the global average.
- Insecurity and adaptation: conflict, economic stress, and new forms of religious practice may preserve belief even as institutions change.
- Measurement: losing a formal affiliation does not necessarily mean losing belief in God.
The resolution challenge
A forecast this distant is only as useful as its measurement plan. The public post defines three directional outcomes—significant increase, no change, and significant decrease—but does not display a numeric threshold for what counts as significant. A formal resolution should specify the survey measure, geographic weighting, comparison baseline, acceptable data sources, and thresholds before the outcome is known.
Survey wording also matters. Asking whether someone believes in God may produce a different answer from asking about a higher power, religious affiliation, prayer, or participation in organized religion. Cross-cultural translation and the availability of global data in 2100 will influence what can be compared reliably with today's baseline.
How to read a 74-year probability
The 50% assigned to decline should be read as a leading scenario, not a confident prediction. It leaves the other half of the distribution on stability or increase because demographic change, political shocks, climate stress, technology, and religious adaptation can reshape the mechanisms that seem strongest today. The long horizon increases both the time for secularization and the number of ways the model can be surprised.
The linked shared report is the updateable record for this question. Any revision should retain its evidence date and explain which new demographic data, surveys, or structural assumptions changed the probabilities. That audit trail is especially important when the horizon is longer than the career of any individual forecaster.
