Analysis · August 2026
In August 2026 the State of the Church in Singapore published its fifth study, and the first designed as a recurring annual survey. It reports 17,151 professions of faith across 122 churches during 2025. Over the five years to 2025, the national census recorded 66,000 fewer residents identifying as Protestant.
Both figures are accurate as reported. They describe different populations, gathered by different methods, and they answer different questions. Placed side by side they do not cancel each other out. They point in the same direction as the cohort analysis on this site.
Sources: State of the Church in Singapore Annual Survey 2026, pp.3 and 9; Singapore Department of Statistics, General Household Survey 2025. Full source list at the foot of this page.
First, the study itself
122 churches, 106,721 regular attendees, nine denominational families, fielded between January and June 2026. It is the only dataset that looks at the Singapore church from the inside at this scale, and there is no substitute for it.
Much of what it reports is useful, and some of it is uncomfortable in ways that deserve a hearing. A fair summary comes first.
Youth aged 13 to 18 fell from 9.8% to 7.3% of attendees since 2022, the sharpest fall of any age band. Senior adults and seniors now make up 38.6%.
Churches classed as Young or Middle-Heavy fell from 34% of the sample to 14%. Balanced rose from 13% to 38%, though 26 of those 46 balanced churches are more than 40% seniors.
17,151 professions of faith during 2025, against 3,348 baptisms. The report notes that the survey never specified what counted as a salvation.
355 long-term overseas missionaries, one for every 301 attendees. Churches above 2,000 send one for every 689. Forty-eight churches sent none.
Average starting salary for a new pastor is $3,161 a month. The national median full-time wage is $5,775, and a starting social worker is on $3,970.
91 attendees per pastoral staff member overall, rising to 167 in churches above 2,000. Cell group participation runs the other way, highest in the largest churches at 77.3%.
Nothing on this page argues against the study. Four theological colleges and a newsroom decided that the Singapore church should be able to see itself, and built the instrument to do it. That is a genuine public good.
What follows concerns how far its findings travel. Several of them are already being read as statements about the Singapore church as a whole. Some carry that weight comfortably. The growth figure carries it least well, and it is the figure most likely to be repeated.
Where it supports the finding on this site
The cohort analysis on this site rests on national household data, which records what people say about themselves. The 2026 survey is the other side of the ledger: what churches count. Different source, different method, same conclusion.
9.8% to 7.3% in four years, the steepest fall of any age band, while the children's share barely moved.
Young and Middle-Heavy churches fell from 34% to 14% of the sample. The shift is statistically significant, and it reverses the 2023 SOTC finding that young churches were getting younger.
Third of three explanations offered: "Church members could have grown up into the older age brackets without new, younger members replacing them."
This matters more than any single percentage. A cohort analysis built on census self-identification can always be answered with the suggestion that people still attend and simply describe themselves differently. The 2026 survey closes that line of argument. The churches themselves are counting fewer young people, and the congregational profile that used to hold most of them has nearly disappeared from the sample.
The part that looks contradictory
If churches added 17,151 people in one year, how did the Protestant population fall by 66,000 across five? The first answer is that the two numbers were never counting the same population, and neither was designed to test the other.
| General Household Survey 2025 | SOTC Annual Survey 2026 | |
|---|---|---|
| Who answers | The individual, about themselves | A church leader, about a congregation |
| Who is sampled | A designed probability sample of resident households | Churches that chose to reply to an emailed invitation |
| Size of the frame | Published | Not reported |
| Response rate | Published | Not reported |
| Key term defined | Religion, from fixed categories | Salvation, left to each respondent |
| Comparable over time | Consistent since 2000 | Sample moved from 144 to 122 churches |
| What it measures | How many people call themselves Protestant | What responding churches report about their own activity |
Sources: SOTC 2026 Annual Survey, pp.2, 3, 9, 10; SingStat GHS 2025 methodology.
The 122 churches account for 106,721 regular attendees. The General Household Survey counts roughly 346,000 residents aged 15 and over who identify as Protestant. Taking the attendance figure at face value, the survey can see something in the order of three in ten of the people the census describes.
106,721 attendees against approximately 346,000 residents identifying as Protestant in GHS 2025 (9.5% of residents aged 15 and over). Illustrative: the survey counts attendance including children, while the census counts identification from age 15, so the true overlap is smaller again.
That gap is not a criticism. No voluntary survey of churches could close it. But it does settle the logic of the comparison. A study covering under a third of the population, made up of churches that chose to respond, is not positioned to revise a national estimate of that population. It describes what happens inside the churches it can see.
The growth figure, examined
Beneath Chart 12, in small print, the report records that "one outlier of church size >2,000 was removed from the final tabulation due to an extremely large number of salvations and baptisms reported." It does not state which church, or how large the excluded figures were. Both follow directly from totals published elsewhere in the same report.
These are not estimates. The report publishes a national total of 17,151 professions of faith and 3,348 baptisms; Chart 11 publishes the same totals with the outlier removed, at 5,088 and 3,052. Subtracting one from the other gives 12,063 and 296 exactly.
On those figures a single congregation, holding roughly one in twenty of the people surveyed, accounts for seven in ten of the study's professions of faith. With it removed, as the charts remove it, the 122 churches report 5,088 professions during 2025 rather than 17,151.
Every input below is printed in the 2026 report. Nothing external is used, and no church is identified.
17,151 professions and 3,348 baptisms. Chart 11 totals, with the outlier removed: 5,088 and 3,052. The differences are 12,063 and 296. No rounding or estimation is involved.
27.8% and 2.9% of 51,643 attendees, which is 14,357 and 1,498. Subtracting Chart 11's band figures of 2,273 and 1,202 leaves 12,084 and 296, agreeing with step 1 to within chart rounding.
7 churches rather than 8, and the overall figure by 121 rather than 122. Band six: 2,273 ÷ 7 = 324.71, printed as 325. Overall: 5,088 ÷ 121 = 42.05, printed as 42.
4.95% and 2.60% once the outlier is removed. Solving each for the denominator that produces it gives 45,919 and 46,231 attendees remaining, so the excluded congregation holds between 5,412 and 5,724 attendees. That places it below the average of its own size band, which runs to roughly 6,600.
Steps 1 to 3 are arithmetic on published totals. Step 4 solves published percentages and therefore carries the rounding in those percentages, which is why it is given as a range. No congregation is identified here, and none should be inferred: the subject is a measurement system, not a ministry.
Measured against its own attendance, it conducted baptisms at roughly 1.8 times the rate of the rest of the sample. On the measure that can be checked independently, it performs unusually well.
The survey did not define what counted as a salvation, and the report says so. A congregation answering an unspecified question may well have answered it accurately on its own terms.
When a single form takes the headline from 5,088 to 17,151, the number describes one organisation's counting convention combined with everyone else's, rather than the state of the Singapore church.
Of all the figures in the report, the growth total is the most likely to travel beyond it, and the most sensitive to this one response.
The research team judged the figure unusual enough to remove from all three growth charts, and disclosed that it had done so. The total quoted in the surrounding text still includes it. A reader moving from the sentence to the chart below it is looking at two different datasets, and nothing in the layout signals the change.
Where the findings travel less well
Each of the five below is visible in the report itself. Together they set the boundary around what the study can support, which matters because several findings are already being quoted well outside it.
Churches were invited by email and replied if they chose. Neither the number invited nor the response rate is reported, so a reader cannot judge how the 122 respondents differ from those who did not reply. Churches with the capacity to compile figures for a research survey are plausibly not a random subset of Singapore's congregations.
SOTC 2026, p.2, survey frame
The 17,151 quoted in the text includes the excluded congregation. Charts 11, 12 and 13, printed immediately below, do not. The exclusion is disclosed in small print under Chart 12, but the two figures appear on the same page without a marker showing which is which.
SOTC 2026, p.9 and Chart 12 footnote
The text states that increases in reported salvations were seen across churches of all sizes. Chart 12 shows the average per church falling in three of the six size bands: under 100 (3 to 2.5), 100 to 250 (9 to 7.7), and 501 to 1,000 (43 to 29.6).
SOTC 2026, p.9, text against Chart 12
The report concludes that there was no statistical correlation between church size and salvations, baptisms or missionary sending. No coefficient, test or significance level is given, so the strength of the conclusion is hard for a reader to weigh. The age-profile finding three pages earlier is reported properly, with a chi-square result. Chart 14 separately shows churches above 2,000 at 688.6 attendees per missionary, against 170 to 230 in every other band.
SOTC 2026, p.22, against Chart 14 and footnote 7
Charts 3 and 4 report an average of each church's percentages, so a 40-member congregation counts as much as a 6,000-member one. Eight churches hold 48.3% of all attendees surveyed. Table 1 then compares those figures with census population proportions, which are counted per person. The two are not the same quantity, and comparisons between them should be read as directional only.
SOTC 2026, Charts 3 and 4, Table 1
A sixth feature is stated openly by the report and is not listed as a gap here: the survey never defined what counted as a salvation, and the report sets out four reasons why professions and baptisms might diverge.
What follows for a reader
The risk in a study of this kind is not error. It is that a reasonable reader draws a conclusion the design was never built to carry, and the direction of that error is not random. Each of the six below runs the same way: towards a more favourable picture than the evidence establishes.
Seven in ten of the reported professions come from one congregation, on a definition the survey never set. With that response removed, the same 122 churches report 5,088 across the year, against a national fall of 66,000 over five.
The conclusion is stated without a test. The report's own charts show churches above 2,000 carrying 167 attendees per pastoral staff member against 91 overall, and 688.6 attendees per missionary against 170 to 230 elsewhere.
Balanced churches rose from 13% to 38% of the sample. The report notes that 26 of those 46 are more than 40% senior adults and seniors. The label describes a spread, not a healthy pipeline.
122 churches replied to an invitation, covering roughly three in ten of the people the census counts as Protestant. Without a published frame or response rate, the distance between the sample and the whole cannot be measured.
It measures arrivals in detail and departures not at all. A congregation can report professions, attendance and event participation every year while the same congregation's twenty-year-olds quietly leave.
The closing word cloud reflects how leaders feel about the year ahead. It is a measure of morale, gathered from the churches that chose to respond, and it is not a forecast.
None of this makes the survey unreliable. It makes it a survey of 122 self-selected congregations reporting their own activity, which is exactly what it says it is on page two.
The difficulty arises when findings from that base are quoted as statements about the Singapore church, and where the gaps in the design happen to run in one direction. A study that counts arrivals but not departures, that averages churches rather than people, and that leaves its central term undefined, will tend to produce a more encouraging picture than a fuller measurement would.
The deeper question
Set aside for a moment whether the growth figure is accurate. A prior question sits underneath it, and it matters more than the arithmetic: is the number of professions of faith a good way to know how a church is doing at all?
Counting is not the problem. A church that counts nothing knows nothing about itself, and the absence of measurement has done its own quiet damage here. The question is which things get counted, because a measure that gets published becomes a measure that gets pursued. Once professions of faith are the figure by which a church describes its year, the activities that produce professions attract attention, budget and Sunday minutes. The slower work that produces formation attracts none of those, because nothing records it and no one asks for it in August.
The argument running through these pages is that a performance-and-evaluation logic, borrowed from the surrounding meritocracy, has taken root inside the church — in what is believed, in how service is valued, in who is trusted with authority, in how a household orders its week. Measurement is where that logic becomes visible at the level of the whole sector rather than the individual congregation. A church that knows itself mainly through counts of arrival will keep producing arrivals, and will keep losing the people it already has, without those two facts ever appearing in the same table.
Scripture measures differently, and slowly. Fruit that lasts. Love that endures. A faith that has been tested and found to hold. None of that yields a figure by December, which is precisely why it loses to anything that does.
Some fell on rocky ground, where it had not much soil, and immediately it sprang up, since it had no depth of soil. But when the sun rose it was scorched, and since it had no root, it withered away.
— Matthew 13:5–6
The parable is not a warning against counting. It is a warning against counting too early. Seed that springs up quickly is the most visible thing in the field and the least reliable sign of a harvest. A survey that records professions in the year they are made is recording germination. Whether anything took root is a question that can only be answered several years later, by going back and looking for the same people.
This is also why the disparity in the 2026 survey is worth sitting with rather than explaining away. The report offers four reasonable accounts of why professions might exceed baptisms, and each of them is plausible on its own. Taken together they describe something the survey cannot see: a large number of people who said yes and are no longer traceable. Whether that gap represents patience, movement between churches, or roots that never went down is the most important open question about the Singapore church, and it is the one question the current design cannot reach.
So the practical question for a congregation is not whether the national figure is right. It is narrower, and more uncomfortable: of the people who professed faith in your church five years ago, how many could you name today, and how many are still here? That question has no dashboard and produces no annual number — which is why the first of the six things a congregation can do is to go and find the answer.
Reading the two together
Suppose the growth figure is broadly right. Suppose Singapore's churches genuinely do add thousands of professions of faith every year. That is the version this site has least difficulty with, because it removes the last comfortable explanation for the decline.
Even on the conservative figure, 5,088 professions of faith in a single year, across churches holding under a third of Singapore's Protestants, the front door is working. The identifying population still fell by 66,000 across five years. What enters is not staying.
A church losing people at the front door would shrink evenly across ages. A church filling at the front and leaking at the back loses whoever is least anchored, and the cohort data on this site shows exactly who that is. The widest losses sit at the transition out of the teens, and again out of the twenties.
Professions of faith, attendance and event participation can all rise in a congregation that is quietly losing its young adults. The 2026 survey measures the first three carefully. It does not yet measure the fourth, and it says so.
"Further studies into church decline, the number of people leaving the Church, would provide insight. Both movement into and out of the Church must be studied longitudinally." That is the survey's own closing observation, arrived at from the opposite direction to the analysis on this site.
Two independent instruments, built by different people for different purposes, now agree on the shape of the problem. Neither was designed to measure retention, and neither currently does. Until something does, the question that matters most about the Singapore church will keep being answered with figures about how many people arrived.
See for yourself
Every figure on this page comes from one of the documents below. Where a number is derived rather than quoted, the derivation is set out in full above and in the data note.
On the Protestant population of roughly 346,000. Derived from GHS 2025: 9.5% of the resident population aged 15 and over. It is used only to scale the coverage comparison, and the coverage bar is illustrative rather than precise.
On the excluded congregation. Its professions and baptisms are exact, being the difference between two totals published in the same report. Its size is given as a range because it is solved from published percentages and carries the rounding in them. No congregation is identified, and none should be inferred.
On the age comparison. The survey's age charts report an average across churches rather than across people. Comparisons between them and census population proportions should be read as directional. The report notes a related caveat itself, at footnote 6.
On scope. Nothing here is a judgement about any congregation's ministry. The subject throughout is a measurement system that Singapore's churches share and rely on.
Spotted an error, or have a source that adds to this? Tell us — corrections are posted openly, not quietly fixed.
Stay with it
New national and church data on Singapore's Christian population arrives through the year, and very little of it gets read alongside the rest. Leave an email address and you will hear when something new is published here. Nothing else, and not often.
Comments
Questions, feedback, or corrections — leave them here.