Does Alzheimer's skip a generation? What a grandparent's diagnosis really means

Short answer: No — Alzheimer's doesn't skip generations, but it looks like it does, and the reason is worth understanding. The rare inherited forms caused by a single gene can't skip: if you inherit the variant, you almost certainly develop the disease, and those forms account for under 5% of all Alzheimer's cases. The common form isn't an on/off switch at all — it's many small genetic effects plus decades of health and circumstance. Risk from that kind of inheritance doesn't skip; it thins out with distance. In the largest study of its kind — 270,818 people with documented family trees — having an affected grandparent, aunt or uncle raised risk measurably, and even affected cousins showed up. And people with no affected relatives at all were slightly below average risk.

This is educational, not medical advice. If Alzheimer's appeared unusually early in your family — under about 65 — that is worth raising with your doctor or a genetic counsellor rather than working out alone.

Why it looks like skipping

Two very different things get called "Alzheimer's," and they inherit in completely different ways.

The rare kind runs in a straight line. Mutations in three genes — APP, PSEN1 and PSEN2 — cause autosomal-dominant familial Alzheimer's, usually with an early onset. If a parent carries one, each child has a 50% chance of inheriting it, and penetrance is high: for PSEN1, essentially everyone who inherits the variant develops the disease; for PSEN2, around 95%. This form cannot skip a generation, because carrying the variant and developing the disease are close to the same thing. It is also rare — these mutations account for less than 5% of all Alzheimer's cases.

The common kind doesn't run in a line at all. For the other 95%+, there is no single switch. Risk comes from many common genetic variants of small effect, plus everything else that happened across a life: blood pressure, hearing, diabetes, education, head injuries, sleep, smoking. You inherit a tendency, in fragments, from both sides — and the fragments get diluted and reshuffled at every generation.

That is why a family can have an affected grandmother, an unaffected mother, and a worried daughter. Nothing skipped. The dice were rolled differently, and the ages at which people happened to die differ too — if a grandmother developed Alzheimer's at 84, her daughter may simply not have reached the age where it would show.

What a grandparent's diagnosis is actually worth, in numbers

Most articles on this question stop at "it's complicated." There is a study that puts numbers on it.

In 2019, researchers used the Utah Population Database — genealogies of Utah's 1800s settler families and their descendants, linked to death certificates — to calculate Alzheimer's risk by exactly how many relatives were affected and how closely related they were. They analysed 270,818 people who had at least 12 of their 14 nearest ancestors documented, among whom 4,436 had Alzheimer's recorded on their death certificate. It was published in Neurology.

Here is roughly what they found, expressed as relative risk — how much higher or lower than the population average.

Parents, siblings and children (first-degree relatives):

(The dip at three and the jump at four reflect very small numbers of families — look at how wide those intervals are. The direction is what's solid, not the precise step at each rung.)

Grandparents, aunts and uncles (second-degree relatives), with no affected parent or sibling: risk was elevated at every count, but only reached statistical significance at three or more affected — where the authors describe roughly a two-fold increase. Fewer than three, and the study couldn't tell an effect from noise.

Cousins and great-grandparents (third-degree relatives), with no closer affected relative: three or more gave RR 1.43 (1.21–1.68). The authors put two or more at a 17–44% increase.

And the result nobody quotes: people with no affected first-, second- or third-degree relatives had RR 0.90 (0.87–0.94) — significantly below the population average. A genuinely clean family history is a mild point in your favour, not merely the absence of bad news.

So: one grandparent with late-onset Alzheimer's is not, by itself, a signal you should be losing sleep over. A pattern of three or more across the wider family is a different conversation. Some pages on this topic tell readers flatly that a grandparent's diagnosis "does not change your risk." That is a simplification the best available data doesn't support.

The honest limitations

This is one very good study, not a settled consensus, and three things about it matter:

• Cases were identified from death certificates, which are known to undercount Alzheimer's. The researchers chose specificity over sensitivity deliberately — so people counted as cases very likely had it, but real cases were missed.
• The second-degree numbers are the weakest, because the records only capture Utah deaths from 1978 onward, which systematically censors the grandparent generation. "Not statistically significant" there means underpowered, not no effect.
• The population was largely of British and Scandinavian ancestry. The authors say explicitly that the estimates "should not be extrapolated to other populations without validation."

The one pattern that changes what you should do

Number of relatives matters. Age of onset matters more.

In the same study, having a first-degree relative who died of Alzheimer's before 70 carried significantly higher risk than one who died later. The under-65 group showed an elevated estimate too — RR 2.24 — but the confidence interval was 0.82–4.88, which crosses 1.0, so it isn't statistically significant. There simply weren't many cases: only 67 of the 4,436 Alzheimer's deaths occurred before age 65, about 1.5%.

Even so, early onset is the recognised clinical signal, and for a straightforward reason: early-onset Alzheimer's is where the single-gene familial forms concentrate. If a parent, sibling or child developed Alzheimer's symptoms before about 65 — and particularly if more than one relative did, across more than one generation — that is worth raising with your doctor and asking about a referral to a genetic counsellor. Not because there's a treatment waiting, but because that pattern is the one where a genetic explanation is genuinely on the table, and where testing decisions deserve proper counselling.

For everyone else — a grandparent in their eighties, an aunt in her late seventies — the family history is one input among many, and not the largest one.

What about the APOE gene?

APOE is the common gene people mean when they say "the Alzheimer's gene," and it doesn't skip generations either — because it isn't a disease gene. It's a risk factor.

For scale: at least one copy of the APOE ε4 variant is found in about 56% of Alzheimer's cases and is associated with roughly three-fold higher risk; two copies appear in about 11% of cases, with eight- to twelve-fold higher risk. Plenty of ε4 carriers never develop dementia, and plenty of non-carriers do.

You inherit one APOE variant from each parent, so it passes down like any ordinary gene — sometimes to you, sometimes to a sibling instead. That ordinary randomness is another reason a family can look like it's skipping when nothing is.

Whether it's worth finding out your genotype is a real decision with real consequences — including some insurance ones that most articles omit. We covered it properly in should I get the Alzheimer's gene test?

What to do with this

If dementia is in your family, the useful move isn't to work out a number. It's to act on the parts of your risk that are actually yours to move.

There's a genuinely encouraging finding here. Pooling three multidomain lifestyle trials — FINGER, J-MINT and MAPT — researchers reported in December 2025 that APOE4 carriers benefited more from structured lifestyle change than non-carriers (interaction p = 0.035). The authors call the result preliminary. But it is the best evidence we have against the fatalism that a family history invites: higher genetic risk did not mean less to gain.

The 2024 Lancet Commission links around 45% of dementia cases worldwide to 14 modifiable factors — hearing, blood pressure, blood sugar, cholesterol, smoking, exercise, sleep, social connection, vision and others. None of them are affected by what your grandmother's death certificate said.

More on this: is dementia hereditary? · is dementia inherited from your mother or father? · can you reduce your dementia risk? · check your dementia risk

How we research and review our content →

This article is educational and is not medical advice, diagnosis, or treatment. The studies cited describe findings in general populations, not any individual. Solenna is not a medical device and does not diagnose, prevent, treat, or cure Alzheimer's disease or any other condition. Risk reduction means lowering probability, not eliminating it; individual results vary and no outcome is guaranteed. If Alzheimer's appeared unusually early in your family, decisions about genetic assessment, referral or testing should be made with a qualified clinician or certified genetic counsellor.