# The Blue Zones case file: what our data check reveals

Saul Newman says the Blue Zones' age records are bookkeeping errors. We recounted 672 deaths in six Sardinian villages: no hoax, no miracle, but from age 90 a trail nobody has explained.

## At a glance

- The twelve centenarians in the six villages are real register entries: 12 plus 1 minus 1 gives 12 again in 2025.
- Of 848 residents lost since 2019, 613 come from deaths exceeding births and only 195 from people moving away.
- From age 70, villagers die 7 percent less often than Sardinians of the same age, but the range reaches zero.
- From age 90, they die 17 percent less often (range 5 to 27 percent), also without centenarians and against neighbouring villages.

Twelve people aged 100 or older live in six mountain villages on Sardinia at the start of 2025, and the six villages together have only 11,217 residents. That works out to 107 centenarians per 100,000 people. Italy as a whole has 38. Numbers like these made the Blue Zones famous: five places in the world where, supposedly, an unusual number of people grow very old.

For a few years now, the story has had a challenger. The demographer Saul Newman argues that many age records come down to poor bookkeeping. Since 2019 the two sides have mostly argued about birth certificates that hardly anyone has seen.

We did something else: we counted. In the six villages and everywhere Newman's theory would have to leave traces. Five annual population counts for all 377 municipalities of Sardinia, broken down by single year of age. 672 deaths in the villages against the rest of the island. The twelve centenarians traced year by year. Plus 248 regions of Europe and 50 years of counts from Okinawa.

The result fits neither side. The villages are not a hoax. They are not a miracle either. And among people aged 90 and over, a trail remains that nobody has explained yet.

Why does anyone doubt the Blue Zones at all?

The doubts come from a single arithmetic trap that almost everyone falls into: the older an age group, the fewer people are left in it. And the more weight every single error in the files carries.

Take an example. Picture 10,000 centenarians. Every year half of them die. After 15 years, on paper, not even one is left, 0.31 of a person to be exact. If that group contains just one 85-year-old whose birth year is off by 15 years in the register, that man suddenly becomes the oldest person in the country. One typo, one record.

Saul Newman of the University of Oxford worked through this mechanism in a peer-reviewed paper in PLOS Biology in 2018. Since 2019 he has gone further in a preprint on the Blue Zones: in Italy, France and England, centenarians cluster precisely where people are poor and die young. His suspicion is that sloppy registers produce records.

How do you spot a guessed age? Round numbers. People who do not know their birth date say "the first" or "the fifteenth". Among the oldest Americans, Newman finds exactly those days piling up: 1.6 times as many birthdays on the first of the month as expected. Real births spread evenly across the month. For Sardinia we cannot run this test, because Istat publishes ages in years only, no birthdays.

For his work on age records he received the 2024 Ig Nobel Prize, the award for research that first makes people laugh and then makes them think.

The preprint has still not been peer reviewed, which means no independent colleague has checked it. His 2018 paper has. Anyone who lumps the two together is cutting a corner.

Japan gives Newman his best-known example. In 2010, police in Tokyo found the mummified body of a man who, according to the register, was 111 years old and had been lying dead in his bed for about 30 years. The review that followed found more than 230,000 register entries for people supposedly over 100 whom nobody could locate.

It had little effect on Okinawa's verified centenarians, though. The study by Saito, Yong and Robine in 2012 showed that these ghost entries sat in a different register from the people whose ages researchers had checked in person.

Why these six villages, of all places?

Because the supporters chose them, not us. In 2004 the Belgian-Sardinian research group around Michel Poulain and Gianni Pes drew a blue circle in the AKEA study around the part of Sardinia where the birth cohorts of 1880 to 1900 most often reached 100. Hence the name Blue Zone.

A paper by Nieddu and colleagues from 2020 names six places as the core of that zone: Arzana, Baunei, Seulo, Talana, Urzulei and Villagrande Strisaili. Five of them sit in the Ogliastra mountains in the east of the island, and Seulo lies just behind them in the Barbagia. The smallest has 765 residents, the largest 3,377.

We adopted that list so that nobody can accuse us of picking villages to suit us. It comes at a price. Without Urzulei, our result would look clearer. We keep it in anyway, because dropping villages after the fact because they spoil the picture is exactly the kind of statistics we are criticising here.

The Italian statistics office Istat publishes two things for every municipality: how many people of each age live there on 1 January and how many people of each age died during the year. From both you can work out, for any age group, how many out of 1,000 people die per year. We did that for the six villages and for the other 371 municipalities in Sardinia.

We did not travel to Sardinia to check birth certificates in parish books. A holiday in the Ogliastra would have been nice. What we could check from a desk, we checked.

Do the twelve centenarians really exist?

Yes, they are in the official files: three men and nine women, spread across all six villages. We also traced these twelve back over five years. At the start of 2024 there were twelve people aged 100 or over, plus one 99-year-old who turned 100 during the year. One centenarian died. Twelve plus one minus one gives the twelve of 2025.

So the count never grows faster than last year's 99-year-olds allow. A register full of ghosts would look different: nobody dies, and the count only ever climbs.

[Image: Istat extract: six municipalities with twelve registered residents aged 100 or over, three men and nine women.]

One of the twelve is even in the newspaper. Luigi Carta from Seulo celebrated his hundredth birthday in July 2023, and La Nuova Sardegna reported it. The Istat file for Seulo lists exactly one man aged 100 or over. That proves he exists. Whether he was really born in 1923, the newspaper never checked in a certificate either. It celebrates what the register says.

So the centenarians are real. The more interesting question is why their share is so high.

Are the villages becoming longer-lived or simply emptier?

Emptier. Between the start of 2019 and the start of 2025 the six villages lost 848 residents, seven percent. Our first thought was the usual story: the young leave, the old stay, the share of old people rises. Only about a quarter of that is true. Only 195 of the 848 are missing because more people moved out than in.

The bulk is simpler. In six years, 365 children were born in the six villages. 978 people died. That leaves 613 fewer residents without anyone booking a removal van. The remaining 40 are corrections in the register.

[Image: Waterfall chart: 12,065 residents in 2019 minus 613 (deaths over births), 195 (migration), 40 (corrections) = 11,217.]

For centenarian statistics this has a consequence. The number of people aged 90 or over rose in those six years from 266 to 296, an increase of eleven percent. Their share of the population, though, rose by 20 percent, from 2.2 to 2.6 percent, because the rest of the population shrank at the same time. Almost half of that rise comes from fewer young people, not from more old ones.

A ranking by centenarians per resident therefore rewards every village that is dying out. Anyone who wants to know whether people there really live longer has to count the deaths.

Do older people there die less often?

Aged 70 and over: not measurably. Aged 90 and over: yes, by about a sixth.

At first we almost wrote the opposite. Out of 1,000 people aged 70 or over, 58 die per year in the six villages and 51 in the rest of Sardinia. The Blue Zone as a death zone?

The catch is the age mix. The over-70s in the villages are on average much older than the over-70s on the rest of the island, because so many 85-, 90- and 95-year-olds live there. Naturally, more of them die.

So we compared 70-year-olds with 70-year-olds, 85-year-olds with 85-year-olds, women with women, each year on its own. Then the picture flips: villagers aged 70 or over die seven percent less often than Sardinians of the same age elsewhere.

Seven percent sounds like a result. It is not. Behind that number are 569 deaths in four years. With so few cases the estimate is imprecise: the true value could be 14 percent fewer, but it could also be one percent more. When "no difference" sits inside that range, we cannot claim a difference. In the chart you see it as a line that touches zero.

[Image: Dot chart: mortality 7 percent lower at 70+ with a range touching zero; 17 percent lower at 90+ with the range below zero.]

For people aged 90 or over the line looks different. Their mortality is 17 percent below that of Sardinians of the same age, with a range of 5 to 27 percent fewer. The whole line stays below zero. Behind it are 218 deaths, still not many. From age 90, a sixth fewer villagers die than in the rest of Sardinia.

We cross-checked this three ways. Against the 19 directly neighbouring municipalities: 16 percent fewer. Against 16 villages nearby with similar size and age structure: 17 percent fewer. And without the centenarians, using only the 90- to 99-year-olds: 18 percent fewer. So the finding does not hang on the twelve whose ages Newman questions.

One warning is due. The 90-plus group was one of several extra questions we put to the same data. Ask enough questions and sooner or later one answer will stand out by chance. The finding is a good reason to keep looking, but not yet a discovery.

Do fake ages leave fingerprints? Four tests

False ages are invisible, but their consequences are not. Newman himself describes the patterns a bad register leaves in the statistics. Four of them we could check without opening a single birth certificate.

First: someone who exists only on paper never dies. If some of the twelve were ghost entries, the death rate above 100 in the villages would have to be strikingly low. It is not. Out of 1,000 centenarians, 455 die per year there and 469 in the rest of Sardinia. Whoever reaches 100 in the Ogliastra dies just as fast as anywhere else on the island.

Second: the shape of the advantage. A wrong age makes someone older on paper, never younger. A person who is really 94 but recorded as 102 dies like a 94-year-old, yet counts among the centenarians.

Such errors therefore pile up in the oldest group, where only twelve real people remain and two false entries already tilt the rate. That is where the advantage would have to be largest. In fact it is moderate among 90- to 94-year-olds, largest among 95- to 99-year-olds and gone from 100 onwards.

[Image: Bar chart: villages have 12 percent fewer deaths at 90 to 94, 26 percent fewer at 95 to 99, 3 percent fewer at 100+.]

That fits real old people who eventually die. But the test is not clean. Behind the figure at 100 and over are only 20 deaths, and the range runs from 38 percent fewer to 51 percent more.

And the shape has a second explanation we do not like: ages padded by a few years, but never up to 100. Because 100 is the line where everyone looks. The mayor comes to congratulate, the newspaper writes, Istat counts centenarians separately. Anyone padding stops short of it. That the advantage breaks off exactly at this line is striking.

The reading does not fully add up, though. Someone who is really 92 but recorded as 97 dies less often than real 97-year-olds. That is exactly why he reaches 100 on paper more often than they do, not less. The advantage should therefore grow from 100 onwards, not vanish. Unless someone corrects the ages downwards at the line.

That would be visible: the villages would then have few centenarians relative to their 90-year-olds. They have more than the island, 4.1 per 100 people over 90 against 2.4. Twenty deaths cannot settle this. This test is a hint with a warning light.

Third: the share of men

Men over 100 are rare everywhere, and Newman treats a high share of men as a warning sign. Up to 94 the villages and the island run alike: about two women per man. Then the lines part. In the rest of Sardinia there are four women per man from 100 onwards, in the six villages only 2.7.

[Image: Line chart: women per man rise from 1.2 at 70 to 4.0 at 100+ in the rest of Sardinia, but only to 2.7 in the six villages.]

The catch: exactly this surplus of men was the core finding of the 2004 AKEA study. Newman reads it as an error, Poulain as a Sardinian trait.

Strikingly, the lines part exactly where the survival advantage is largest: from 95. The same spot where test 2 has its warning light.

Split by sex, women aged 95 to 99 in the villages die 31 percent less often, while for men the range on 26 deaths is too wide to say anything. Behind the figure at 100 and over are three men. Whether they survive better or are older on paper, our data cannot settle. This test stays open.

Are there more centenarians where people are poor?

Fourth: poverty. Newman's strongest argument is that record ages cluster where people are poor and die young. We traced this for 248 regions of Europe using Eurostat figures for 2024, with poverty values available for 220 of them. Record age here means centenarians per 100,000 residents. You can see the result in the two scatter plots.

[Image: Two scatter plots of 248 European regions: centenarians rise with life expectancy at 65, not with poverty risk.]

In both pictures every region is a dot. The vertical axis always shows the same thing: the centenarians. On the left, the horizontal axis asks "die young?": how many years does a 65-year-old in this region still have ahead on average? In Bulgaria 16 to 17, in Madrid 23. So 65 is not a record age but the starting point from which a region's ageing is measured.

Why from 65 and not from birth? Life expectancy at birth hangs on infant mortality, accidents and drugs. From 65 it measures ageing only, the part of life that centenarians are about.

Newman does something similar: for Italy he takes survival to 55 as a marker of midlife and compares it with survival to 100. That centenarians track life expectancy at 65 is no surprise but the expectation. Both figures measure the same thing: how well old people survive.

Why not 1.0?

That it is 0.82 and not 1.0 has two reasons. Life expectancy at 65 depends mostly on deaths between 65 and 90, centenarians on those after. And today's centenarians were born in 1924, while life expectancy measures today's death rates.

Why run the test at all? Because Newman's claim breaks exactly this expectation: in his data, record ages cluster where people die young. That would be the paradox that gives bad registers away. Regions far above the diagonal would be the suspicious ones.

The dots form a clear diagonal: where 65-year-olds live longer, there are more centenarians. The rank correlation is 0.82, a very tight relationship. On the right, the axis asks "poor?": the same centenarian figure against the share of people Eurostat counts as at risk of poverty or social exclusion. The cloud is a cloud. Correlation 0.06, practically none.

In Europe today, centenarians appear where people live long, not where they are poor. Newman's data are older and sorted by person rather than by region. That narrows his claim, it does not refute it.

No test convicts the registers, none acquits them. The clearest is the map of Europe: poverty produces no centenarians today. The shape of the advantage and the share of men remain hints with a warning light. None replaces a look at the 1925 baptismal register in Villagrande.

What about Okinawa?

Okinawa is a Blue Zone disappearing before our eyes. In 1975 the Japanese island had 5.5 times the centenarian rate of Japan as a whole. In 1990 it was 4.8 times, in 2025 0.96 times. Okinawa today sits below the Japanese average. Among all 47 prefectures the island ranked first in 2008 and 37th in 2025.

[Image: Line chart, Okinawa's centenarian rate over Japan's, 1975 to 2025: 5.5 times, dip 3.9 (1980), 4.8 (1990), 0.96 (2025).]

Newman explains it this way: the early numbers were inflated because the family registers burned in the war in 1945 and were rebuilt from oral statements. The defenders say the mainland caught up while Okinawa adapted its diet to the American occupation after 1945.

The curve does not settle this, since it falls smoothly enough for both explanations. It rules out only one version: a one-off register clean-up, say after the mummy scandal of 2010, would have left a kink. There is none.

Nicoya in Costa Rica shows something similar. Luis Rosero-Bixby recalculated in 2023 with register data from 1990 to 2020: men born in 1905 died a third less often there than in the rest of Costa Rica. Men born in 1945 die ten percent more often. The advantage held for one generation. Their grandchildren no longer have it.

What does this mean for you?

Less than the Blue Zones books promise. Our numbers say that in six Sardinian villages people aged 90 and over die less often than elsewhere. They do not say why. It could be genes, mountains and work, health care, chance, or who stays in old age and who leaves. What these people ate is not in any Istat file.

For questions about food there are better sources than maps. The PREDIMED trial followed 7,447 Spaniards at high cardiovascular risk for five years and showed that a Mediterranean diet with olive oil or nuts made heart attacks and strokes about 30 percent rarer. That is a result about one specific habit. A photo of a smiling centenarian is not.

What we found fits in three sentences. The Sardinian centenarians are real, and the villages are no forgery. Their famous share comes largely from the villages dying out. And from age 90, a sixth fewer people die there than elsewhere, for reasons nobody knows.

The deaths of 2025 and 2026 will show whether this trail holds, once Istat publishes them. We will run the numbers again.

How we did the maths

All data are public and come from the Italian statistics office: residents by age on 1 January for 2021 to 2025, the provisional deaths by municipality, age and sex for 2021 to 2024 and the municipal balances for 2019 to 2024. The years 2021 to 2024 are the last for which both are complete.

Someone who dies in March has lived only three months of that year. So we count lived time, not heads: the mean of the resident counts on 1 January of two consecutive years. Istat computes its own death rates the same way.

We compared using the Mantel-Haenszel method, which keeps five-year age groups, sex and calendar year separate and compares death rates only within each of those groups.

Which comparison groups and age groups we would test was fixed before the calculation: the rest of Sardinia, the 19 neighbouring municipalities according to the official municipal boundaries and 16 places within 50 kilometres with half to double the population and a similar share of older residents.

The calculation without centenarians came only after we had seen the result and is labelled accordingly.

We checked our totals against a second Istat file with monthly values: all 12,064 cells matched.

The provisional deaths for the six villages differ from the final municipal balances by two cases out of 672. The death rates above 100 and by age band are crude rates without the age adjustment. The European figures come from Eurostat for 2024, the Okinawa series from the September counts of the Japanese health ministry.

The question is not new. Caselli and Lipsi compared Sardinian mortality back in 2006, De Santis and colleagues did so in 2026 with municipal data from 2002 to 2018. A preprint by Olshansky and colleagues has asked since January 2026 whether the Blue Zones are disappearing. Our four years add to that work. They are not a peer-reviewed study.

## Bottom line

**What is certain** The twelve centenarians are real register entries, traceable year by year. Their high share comes mainly from the villages shrinking: 613 of the 848 lost residents are missing because more people die than are born.

**What is open** From age 90, villagers die a sixth less often than Sardinians of the same age, also against neighbouring villages. Whether that is better survival or padded ages, 218 deaths cannot decide. The deaths of 2025 and 2026 will tell.

**What it means for you** No recipe. Death registers contain no meal plans. To learn what a diet does, read trials about the habit itself, such as PREDIMED, not the address of a centenarian.

## FAQ

### Are the Blue Zones real or fake?

The Sardinian centenarians are real register entries: twelve people in six villages, traceable year by year. Whether every birth year is correct, nobody can tell from a desk. Above 100 the villagers die as fast as the rest of Sardinia, though on only 20 deaths.

### Why does Saul Newman doubt the age records?

In the highest ages so few people remain that a single wrong birth year can create a record. His 2018 peer-reviewed paper works through this mechanism, and his Blue Zones preprint adds that record ages cluster in poor regions with bad registers. Our European check found no link with poverty today.

### Do people in the Sardinian Blue Zone really live longer?

From age 70, not measurably: seven percent lower mortality with a range that reaches zero. From age 90, yes: about 17 percent lower against three comparison groups, based on 218 deaths in 2021 to 2024. The reason is unknown.

### Does the Blue Zones diet make you live to 100?

Nothing in our data says so. Municipal death counts contain no meal plans. Trials such as PREDIMED show that a Mediterranean diet cuts heart attacks and strokes by about 30 percent in people at high risk, which is a different claim from reaching 100.

## Sources

- https://doi.org/10.1016/j.exger.2004.06.016
- https://doi.org/10.3390/nu12061621
- https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.2006776
- https://doi.org/10.1101/704080
- https://doi.org/10.4054/DemRes.2012.26.11
- https://onlinelibrary.wiley.com/doi/10.1111/joim.13764
- https://demo.istat.it/app/?i=POS&l=it
- https://demo.istat.it/app/?i=P02&l=it
- https://www.istat.it/notizia/dati-di-mortalita-cosa-produce-listat/
- https://www.istat.it/storage/dati_mortalita/giugno-2026/decessi-comunali-giornalieri-provvisori-4.zip
- https://www.istat.it/storage/dati_mortalita/giugno-2026/decessi-comunali-mese-provvisori-3.zip
- https://www.istat.it/storage/cartografia/confini_amministrativi/generalizzati/2025/Limiti01012025_g.zip
- https://pmc.ncbi.nlm.nih.gov/articles/PMC3288408/
- https://ec.europa.eu/eurostat/data/database
- https://uub.jp/pdr/j/hyakusai.html
- https://www.demographic-research.org/articles/volume/49/27
- https://doi.org/10.1056/NEJMoa1800389
- https://www.demographic-research.org/articles/volume/14/13
- https://link.springer.com/article/10.1186/s41118-025-00262-3
- https://doi.org/10.21203/rs.3.rs-8497718/v1

---

_Canonical: https://longevity-germany.com/en/articles/blue-zones-inspected · Part of Longevity Cities · Updated 2026-09-16_