Mormonism Explained
Have 40% of Mormons Left the Church?
2.17 million people left the Church’s records in 25 years. Almost all of them died. A widely-cited book puts the figure at 40% of once-active members. We ran the Church’s own membership ledger against it — and found the number lives in a gap between two measuring systems that nobody has named.
The Number Everyone Is Quoting
Forty percent. It is the figure that moves through Latter-day Saint conversation now — in podcasts, in bookstores, in Sunday School classes, in text threads between worried parents.
It comes from Torn: Why People We Love Are Leaving the Church and What We Can Learn from Them (2026), by Jeff Strong, a former bishop, mission president, and BYU faculty member. The book carries a foreword by Steve Young. It draws on a survey of more than 20,000 current and former members — which Strong describes as the largest study of Latter-day Saint disaffiliation ever conducted.
Strong’s claim is specific: roughly 40 percent of Latter-day Saints who were active in 2000 have stepped away from Church participation in the twenty-five years since.
If four in ten once-active members have gone, that is not a trend. That is a rupture.
So we did something that, as far as we can determine, nobody else in this debate has done. We went to the Church’s own books.
Every April, in general conference, the Church publishes a statistical report. It lists total membership. It lists how many children were added to the records that year. It lists how many converts were baptized. Twenty-six consecutive years of those reports are public — and every one of them is reproduced in the appendix below.
Those numbers can be made to answer a question they were never designed to answer — if you are willing to be honest about what they cannot tell you.
What We Found
Between the end of 2000 and the end of 2025, 2,170,605 people left the records of The Church of Jesus Christ of Latter-day Saints. Under every defensible assumption we tested, the large majority of them left the same way every human being eventually leaves every organization.
They died.
How the Ledger Works
The arithmetic is simple enough to check on paper. Membership at the end of any year equals membership at the end of the prior year, plus children of record added, plus converts baptized, minus everyone who left the rolls for any reason.
The Church publishes every term in that equation except the last one. So we solved for it.
Twenty-five years, 2001–2025
People
Twenty-five years, 2001–2025
People
Twenty-five years, 2001–2025
People
Twenty-five years, 2001–2025
People
Twenty-five years, 2001–2025
People
Source: annual statistical reports of The Church of Jesus Christ of Latter-day Saints, 2000–2025. Membership rose from 11,068,861 to 17,887,212 over the period.
That gap contains everything. Deaths. Voluntary resignations. Withdrawal of membership through a membership council. Unbaptized children removed from the records at age nine. Administrative clean-up of the address-unknown file.
Why this cannot be a single number
The Church does not publish deaths. So the only way to estimate departures is to estimate mortality and subtract. That makes the result a band, not a figure — and we publish it as a band.
Single years are worse still: several turn negative under higher mortality assumptions, because members whose deaths go unrecorded stay on the rolls until age 110. We never publish a single-year residual, and neither should anyone citing this work.
Almost All of Them Died
We ran the model at three mortality assumptions: four, five, and six deaths per thousand members per year. The spread between them is the honest measure of our uncertainty.
~350,000
Central estimate of all non-death removals from Church records, 2001–2025 — approximately 14,000 people per year, out of a membership that grew past 17 million.
Put that in proportion — and we will give you both denominators, since they differ. The ceiling, 715,000, equals about four percent of current membership; the central estimate, 350,000, is roughly 1.7 percent of the ~20 million people who were ever on the rolls during the period (11.07 million at the start plus 8.99 million added). Strong’s forty percent, applied to the active membership of 2000, would imply somewhere between 1.3 and 1.8 million departures.
There was no mass exodus. There was mortality.
Three Independent Checks
A model is only as good as its assumptions, so we anchored the mortality band against outside evidence — including evidence we did not generate.
The Church Said It First
In 2016, Church spokesman Eric Hawkins told the Salt Lake Tribune that name-removal requests had run below one-tenth of one percent of membership annually — fewer than 1 in 1,000 members — for more than twenty consecutive years. On 2015 membership, that is under 15,600 people. Our central estimate: roughly 14,000 per year. The Church’s own public statement lands almost exactly on our independently derived figure.
An Outside Statistician
Biostatistician Christian Anderson, working from the same published reports, estimated roughly 30,000 resignations and membership withdrawals in 2016 — about 150,000 across four years. He described it as an order-of-magnitude estimate. It matches the ceiling of our band.
The Demographers
Matt Martinich, who has tracked Church growth data for two decades, has repeatedly described the annual gap as stable at 90,000 to 120,000 across the last decade. Independent analysis put total attrition at 0.86 percent of the membership base in 2024, against 0.87 percent in 2018.
Back-solve that: a gap of 90,000 to 120,000, minus externally anchored removals of 15,000 to 30,000, implies recorded deaths of roughly 75,000 to 105,000 per year. On a base of 14 to 17 million, that is 4.9 to 6.6 per thousand — inside our band.
But Rolls Are Not People in Pews
Here is where an honest investigation has to stop and turn on itself.
Everything above measures one thing: movement on and off a membership record. It is a complete census of that. It is also, for the question most people actually care about, close to useless.
Because a Latter-day Saint who stops attending does not vanish from the rolls. There is no automatic removal for inactivity. Someone can stop coming in 2003 and still be counted in the 17.9 million on the most recent report.
The ledger
Counts membership records. Complete. Cannot see whether anyone is in the room.
The survey
Counts participation and identity. Can see the room. Cannot see the records.
Strong’s forty percent is a participation claim. Our 2.17 million is a records claim. They are measurements of different things, taken with different instruments, and neither one can confirm or refute the other.
This is the part almost nobody in the argument has said out loud — including, in our reading, both Strong and several of his critics.
Which means the ledger cannot end this. It can only establish what did not happen: there was no mass formal exodus. What happened at the level of the pew requires a different instrument entirely.
So We Checked the Surveys
The best instrument for this question is the Pew Research Center’s Religious Landscape Study, which recruits by mailing invitations to randomly selected households drawn from a postal database covering roughly 90 percent of American homes. It has been fielded three times.
The trend is real, and we are not going to soften it. Retention has fallen from about 70 percent in 2007, to 64 percent in 2014, to 54 percent in 2023–24.
Though the Endpoint Is Probably Too Low
Stephen Cranney has flagged a problem in the most recent wave. Between surveys, the share of Americans reporting a Latter-day Saint upbringing jumped from 1.7 percent to 2.1 percent — implying more people raised in the faith than total membership can account for. The entire increase landed among those who no longer identify. Meanwhile the share currently identifying as Latter-day Saint held flat at 1.1 percent in both waves.
That caveat travels with the trend wherever we cite it.
Where That Ranks
Using the 2024 Pew data, a Brigham Young University research team found Latter-day Saint retention higher than every other Christian denomination except Orthodox Christians and Catholics, and statistically indistinguishable from Catholics and Baptists. On the stricter measure — still identifying and attending at least monthly — Latter-day Saints ranked highest of any group measured, at 42 percent.
For scale: 74 percent of Americans raised with no religion remain unaffiliated as adults. Nationally, leaving is the sticky category.
Identify Is Not Attend — and Neither Is “Left”
One distinction dissolves half the alarming headlines. Fifty-four percent still identify. Forty-two percent still identify and attend at least monthly. The twelve points between them have not left the Church — they are less active.
Read the population raised Latter-day Saint as three groups, not two: 46 percent no longer identify, 12 percent identify but attend less than monthly, and 42 percent identify and actively practice — the highest active-retention rate of any group measured.
Identify, attend, and “left” are three different measurements. Keep them straight, and half the scary headlines dissolve.
The Number We Will Not Use
One major instrument disagrees sharply, and we would rather surface it than have someone find it later.
The Cooperative Election Study, run out of Harvard, estimates a 45 percent decline in the Latter-day Saint share of the U.S. population from 2014 to 2025. Pew puts the comparable decline at 8 percent. PRRI puts it at 16 percent. Neither of the latter two is statistically significant.
Cranney and Josh Coates argue that when a politics-focused omnibus survey and purpose-built religion surveys disagree about religion, the religion surveys should govern. We adopt that reasoning — and we are telling you the other number exists.
On Torn’s own data
Strong’s survey was distributed through online Latter-day Saint communities, including disaffiliated ones, and shared onward within them. It was long enough that only motivated respondents finished. Cranney and Coates found its results diverge sharply from representative samples.
Asked about this directly by the Deseret News in May 2026: did the opt-in, non-representative sample shape his conclusions? His answer: “Yes, it does. As a non-probability sample, this cannot be generalized.” He added that the findings should be read as descriptive rather than as precise population estimates.
He also defended the approach, and the defense is fair: large mixed-method studies are common in the social sciences, and selection bias does not automatically make data useless. We take that point seriously.
The author says his data cannot be generalized to the population. “Forty percent of Latter-day Saints” is a population estimate.
That is not a knock on Strong, who has been consistent about this when asked. It is a problem with how the number has travelled — repeated as a measurement by people who did not read the interview.
What We Are Not Saying
We are not saying nobody leaves. Retention is softening, and the surveys are clear about it.
We are not saying Strong is wrong about culture, kindness, or listening. On that, he is right — and so were President Uchtdorf in 2013, Elder Ballard in 2016, and President Nelson in 2023, who said much the same before this book existed.
We are saying the forty percent figure has not been measured by any instrument capable of measuring it.
Appendix — The Full 25-Year Ledger
Every input year, as published in the Church’s annual statistical reports, with the departure gap computed per year: prior-year membership + children of record + converts − reported membership. Single-year gaps bundle deaths with all removals and are never cited alone.
| Year | Membership | Children of record | Converts | Gap (all departures) |
|---|---|---|---|---|
| 2001 | 11,394,522 | 69,522 | 292,612 | 36,473 |
| 2002 | 11,721,548 | 81,132 | 283,138 | 37,244 |
| 2003 | 11,985,254 | 99,457 | 242,923 | 78,674 |
| 2004 | 12,275,822 | 98,870 | 241,239 | 49,541 |
| 2005 | 12,560,869 | 93,150 | 243,108 | 51,211 |
| 2006 | 12,868,606 | 94,006 | 272,845 | 59,114 |
| 2007 | 13,193,999 | 93,698 | 279,218 | 47,523 |
| 2008 | 13,508,509 | 123,502 | 265,593 | 74,585 |
| 2009 | 13,824,854 | 119,722 | 280,106 | 83,483 |
| 2010 | 14,131,467 | 120,528 | 272,814 | 86,729 |
| 2011 | 14,441,346 | 119,917 | 281,312 | 91,350 |
| 2012 | 14,782,473 | 122,273 | 272,330 | 53,476 |
| 2013 | 15,082,028 | 115,486 | 282,945 | 98,876 |
| 2014 | 15,372,337 | 116,409 | 296,803 | 122,903 |
| 2015 | 15,634,199 | 114,550 | 257,402 | 110,090 |
| 2016 | 15,882,417 | 109,246 | 240,131 | 101,159 |
| 2017 | 16,118,169 | 106,771 | 233,729 | 104,748 |
| 2018 | 16,313,735 | 102,102 | 234,332 | 140,868 |
| 2019 | 16,565,036 | 94,266 | 248,835 | 91,800 |
| 2020 | 16,663,663 | 65,440 | 125,930 | 92,743 |
| 2021 | 16,805,400 | 89,069 | 168,283 | 115,615 |
| 2022 | 17,002,461 | 89,059 | 212,172 | 104,170 |
| 2023 | 17,255,394 | 93,594 | 251,763 | 92,424 |
| 2024 | 17,509,781 | 91,617 | 308,682 | 145,912 |
| 2025 | 17,887,212 | 91,835 | 385,490 | 99,894 |
| Totals 2001–2025 | +6,818,351 net | 2,442,033 | 6,546,923 | 2,170,605 |
Methodology
Model. For each year: Membership(t) = Membership(t−1) + Children of Record(t) + Converts(t) − Deaths(t) − Removals(t). All terms but the last two are published. Deaths are modeled as a crude rate on prior-year membership across a 4–6 per 1,000 band; Removals is the residual.
What the residual contains. Voluntary resignations, membership-council withdrawals, unbaptized children of record removed at age nine, address-unknown purges, and any error in the mortality assumption. It is an upper bound on voluntary departures, never a count of them.
Why the recorded death rate runs below actuarial expectation. Members whose deaths are never reported remain on the records until age 110. Commentary on the 2024 report estimated true mortality for a population this size at 90,000–134,000 annually. If deaths are undercounted, our removals band is too high, not too low.
Data provenance. Figures for 2018–2025 were verified against published sources. Figures for 2000–2017 were drawn from the compiled statistical-report archive and eye-verified against original conference reports before publication.
Reproduce it. The full year-by-year input table and sensitivity output are published in the companion technical report. We would rather be corrected than cited uncritically.
Verification Stamps Applied
[VERIFIED] 2.17M gap · Hawkins statement · Pew trend · BYU Studies rankings
[MODELED] Death band · removals band · all derived figures
[CONTESTED]CES divergence · Strong’s 40% (no instrument confirms or refutes)
[DATA GAP] Mechanism attribution within the 2024 record gap