A life-saving transplant can depend on finding one person.

Blood begins here.

Inside your bones is marrow — the tissue that makes blood. In a healthy adult it turns out more than 300 billion new red cells, platelets and neutrophils every day.

None of that is growth. It is replacement. Blood cells wear out constantly, and the marrow never stops.

All of it starts as one kind of cell.

Blood-forming stem cells are rare — a tiny fraction of what is in marrow — and every red cell, white cell and platelet in you descends from them.

They do two things nothing else does. They copy themselves, which is how the supply keeps going for a lifetime. And they can become whichever blood cell the body asks for.

This is the cell a donor gives. Not an organ. The cell that starts blood.

Three jobs, and no time off.

Early on, each stem cell takes one of two roads — myeloid or lymphoid. That single fork is why leukemia later comes in myeloid and lymphoid versions.

What the marrow is making, and how long it lasts

The two-road branch is a teaching simplification — current research describes blood cell development as more of a continuum than a clean tree. What holds is the tempo: because blood turns over this fast, a marrow that stops working properly shows up in weeks, not years. Acute leukemia is called acute because the cells stop maturing — and that turnover is why the shortage is felt almost at once.

Leukemia begins with a cell that never grows up.

A blast is an ordinary immature blood cell — an apprentice, on its way to becoming something useful. In leukemia the apprentice stops developing, keeps dividing, and stops dying on schedule.

Three rules break at once. Do not mature. Do not stop copying. Do not die. Nothing invades from outside; the system that makes blood turns on itself.

This is not something a person brings on themselves. Leukemia is not contagious, is only rarely inherited, and in most people no cause is ever found.

The damage is done by crowding.

Leukemia cells fill the marrow and block healthy blood cells from developing. Much of the harm is not what those cells do — it is what the marrow can no longer make.

Which means the symptoms of leukemia are the absence of the three things you just met: red cells, platelets, and white cells that work.

From the inside, it feels like an ordinary bad month.

Losing each of the three costs something different, and none of it announces itself as cancer.

There is a cruelty in the arithmetic, too. The white cell count can be normal, low, or very high — and when it is high, those cells are still too immature to fight anything. More cells, less defense.

What that absence feels like

• Too few red cells — tiredness a full night of sleep does not touch, breathlessness on the stairs, a paler face than usual.
• Too few platelets — bruises you cannot account for, nosebleeds, bleeding gums, and sometimes petechiae: a pinprick rash that does not fade when you press a glass against it.
• Too few working white cells — infections that keep returning, or one that will not clear.
• And something else again: bone pain, and in children sometimes a limp, from a marrow so crowded that it presses outward on the bone.

Every one of these has ordinary explanations, and most people — most children — with one or more of them do not have leukemia. If any of it persists, the answer is a doctor, not this page. But the disguise is also why acute myeloid leukemia is so often found late: of AML diagnoses in England in 2019, about 5% came through an urgent suspected-cancer referral, while around 6 in 10 arrived through an emergency admission.

What finds it is an ordinary blood test.

A full blood count counts what is there and finds the numbers wrong. A drop of blood on a glass slide can show blasts circulating where they should not be. The same test people have for insurance physicals and unexplained tiredness is the test that finds leukemia.

Confirming it means taking marrow, usually from the back of the hip. In adults it is usually done awake, under local anesthetic that burns for a few seconds, and patients consistently warn each other about the same thing: a pulling sensation that can travel down the leg. About thirty minutes, then a small bandage and no shower for a day. Children are almost always given sedation or a general anesthetic.

Then the marrow is read twice over — flow cytometry asks what the cell is, and genetic testing asks what went wrong. Those genetic results sort the leukemia into favorable, intermediate or adverse risk, and that sorting is a major input into whether anyone starts looking for a donor. An abnormal blood count on its own is not a diagnosis.

Four leukemias. Two questions.

How fast is it moving — acute or chronic? And which road did the cell take — myeloid or lymphoid? Cross those two questions and you get the four main types: AML, ALL, CML and CLL.

The differences are not small. Some people with CLL never need treatment at all. CML is now usually held in check by a daily tablet aimed at the exact genetic fault driving it. ALL is the most common cancer in children, and in the United States five-year survival for children under 15 with ALL has risen from about 60% for those diagnosed in 1975 to about 90% for those diagnosed in recent years.

So the word “leukemia” on its own tells you very little about what someone is facing. Which leukemia is not a detail. These are US population figures covering people diagnosed years ago; they lag behind current treatment, they differ between countries and health systems, and they cannot predict any one person’s outcome.

Chemotherapy is not a smart bomb.

It kills cells that divide quickly. Leukemia cells divide quickly, so they are hit hardest — but so is the lining of the mouth and gut, the hair follicles, and the marrow itself.

For AML the most common intensive start is nicknamed 7+3: one drug running continuously for seven days, another given on the first three. It is not the only regimen, and it is not right for everyone. Among adults under 60 who are well enough for it, roughly 60 to 70% reach a first remission; for older adults, who are the majority of people diagnosed with AML, the figure is lower. Acute lymphoblastic leukemia runs the opposite way — a course of two to three years, most of it low-dose maintenance taken at home.

What people describe

• Mouth and gut ulceration. In a 1999 interview study, 38 adults who had been through a transplant were asked to name the single most debilitating side effect: 42% chose mouth sores, 13% nausea and vomiting.
• Hair, usually starting two to four weeks in, with the scalp aching while it goes. It grows back — often finer, sometimes curlier, sometimes a different color.
• Fatigue that sleep does not fix, and that people describe as different in kind from ordinary tiredness — less like being tired than like having had the energy taken out of them.
• Food that tastes of metal, or of nothing at all, so people eat with wooden cutlery and stop eating out of pointlessness rather than nausea.
• A temperature over 38°C (100.4°F) that is no longer a nuisance but an emergency. Many services aim to start antibiotics within an hour of arrival.

Nausea is far better controlled than it once was — medicines are given before treatment rather than after symptoms start — though control is not complete. Regimens differ by subtype, age and country, and are chosen by a treating team.

Treatment can push it back. It does not always stay back.

For many people chemotherapy brings the marrow back within normal limits and clears blasts from the blood. That is remission — and remission is not the same as cure.

Sensitive testing can still find traces that a microscope cannot. When those traces remain, when the leukemia’s genetics mark it as high risk, or when it returns after treatment, the conversation changes.

That is the point at which a team may begin looking for a donor.

Most people with leukemia never need a donor.

It is worth saying plainly, because a page like this can imply otherwise. Chemotherapy alone cures most children with acute lymphoblastic leukemia, and a minority of adults with acute myeloid leukemia. Some people with CLL are never treated at all.

Chronic myeloid leukemia is now usually held in check by a daily tablet aimed at the exact genetic fault driving it. The European LeukemiaNet 2020 recommendations report that for someone newly diagnosed in chronic phase, life expectancy is now very close to that of age-matched people in the general population — at least in Western countries, where those drugs are reliably available. And one subtype of AML, acute promyelocytic leukemia, can often be treated with little or no conventional chemotherapy: in the Italian-German APL0406 trial, all 77 evaluable adults with non-high-risk disease given a vitamin A derivative with arsenic trioxide reached complete remission. APL is still treated as an emergency, because untreated it can bleed catastrophically within days.

A transplant is offered when the risk of the leukemia returning outweighs the risk of the transplant itself. That judgment belongs to a patient and their team.

When it is the path, the marrow has to be emptied first.

Conditioning takes about a week or two. St. Jude, explaining it to the families of children, puts its three jobs plainly: make room for the donor cells to grow, weaken the immune system so the body accepts them, and kill any cancer cells left behind.

Where radiation is part of it, it is given to the whole body: about twenty minutes a side, up to ninety minutes a session, once or twice a day for one to four days. Nothing is felt while it runs. What follows can include nausea, mouth sores and fatigue, and years later infertility and a raised risk of a second cancer. The line families most need to hear is that afterward there is no radiation left in the patient or on their clothes.

This is the genuinely dangerous part. For a stretch of days the patient has almost no working marrow at all, deliberately.

Then a bag of cells, through a line, while you are awake.

The transplant is not surgery. The cells run in through a central line already in place, over anything from half an hour to a few hours, and it looks like a blood transfusion. Nigel, a transplant recipient, told Anthony Nolan: “To be honest it’s just a normal day, it’s a transfusion just like all the ones I’d had before.”

That day is renumbered day zero. Everything before it is counted in minus days and everything after in plus days. Donor cells find their way into the marrow on their own — nobody places them there.

Many recipients count a second birthday from day zero for the rest of their lives.

Then weeks of waiting for a number to go up.

Blood is drawn every morning. Until the donated cells take hold the patient has almost no working immune defenses, and lives on transfused red cells and platelets from other donors entirely.

What that stretch looks like

• A single room with the door kept shut, one or two named visitors, and no fresh flowers on the ward — rules differ between hospitals, but the shape is the same: the room protects you by keeping everyone out of it.
• Someone else’s blood, constantly. Red cells keep for roughly 35 to 42 days and platelets for only about 5 to 7, depending on country and processing, so the need never stops.
• A care partner on the clock around the clock. Transplant centers generally require one to be named before transplant and available twenty-four hours a day for roughly the first hundred days, and studies of transplant caregivers find their distress runs as high as the patients’ own.
• The fertility conversation, held within days of diagnosis and often with someone in their teens or twenties. Collecting eggs needs roughly two weeks of stimulation first, and acute leukemia often cannot wait that long; banking sperm takes a day, and ovarian tissue can sometimes be frozen instead.

Engraftment is the first morning the count rises instead of falling. Families learn to read a blood panel the way other people read a weather forecast.

And then, for most people, an ordinary year.

The first hundred days are the closest watched. After that comes a slower rebuild: months of immunosuppression, an immune system that takes something like one to two years to come back, and childhood vaccinations given again from scratch because the old immunity left with the old marrow.

Going back to work or school often means six months to a year, and people commonly build up an hour at a time. Long-term follow-up continues for life, because treatment this heavy leaves things to watch for.

If this is happening to you or someone you love, the people to ask are your own transplant team — they know your case, and this page cannot. Blood cancer charities in most countries run free helplines staffed by nurses.

What arrives is not just cells. It is an immune system.

This is what makes a donor transplant different from any drug. The donor’s immune cells rebuild the blood — and they can also recognize leukemia cells that survived treatment and attack them. That effect is called graft-versus-leukemia, and it is a major reason a donor transplant can work where chemotherapy alone has not.

The same power has a mirror image. If the new immune system reads the patient’s own tissue as foreign, it attacks that too — graft-versus-host disease. An acute form develops in roughly a third to a half of donor transplant recipients, most often in the skin, gut and liver; a chronic form can follow later and reach the eyes, mouth, lungs and joints. It is serious, sometimes life-threatening, and teams work hard to prevent it. Rates vary widely with donor type and prevention regimen, and have been falling.

So what a donor gives is not spare parts. It is a living immune system, and it has to recognize the body it lands in.

And it does not end at engraftment.

A transplant can work and the leukemia can still come back. Relapse is why teams keep watching for years afterward, and it is the hardest thing to hold in a story like this one: the morning the count finally rises is not the morning it is over.

The chronic form of graft-versus-host disease can begin months later and settle in the eyes, mouth, skin, lungs or joints. Conditioning leaves its own long shadow — infertility, thyroid and hormone changes, and a raised risk of a second cancer years on. Long-term follow-up clinics exist because these things are known and looked for, not because they are expected.

None of that is a forecast. Risk differs enormously with age, disease, regimen and donor, it has been falling, and a great many people transplanted years ago are living ordinary lives now. It is here because a version of this story that stopped at the good morning would be a recruitment page rather than an explanation — and because the people living through the year afterward deserve to see it written down.

A match is more than a blood type.

The markers being compared are called human leukocyte antigens — HLA. They sit on the surface of most of the cells in your body, and they are how your immune system tells your own cells from cells that do not belong.

So this is not about blood group. It is about whether an immune system built by someone else will read your body as home.

They are the most variable genes in the human genome.

As of release 3.65 of the IPD-IMGT/HLA Database, published in July 2026, 45,421 different versions had been catalogued — and the count grows every quarter. One gene alone, HLA-B, accounts for more than 11,000 of them.

But raw variety is only half the story. What actually decides whether you match is how common your particular inherited combination is — and that depends on which populations your ancestors came from.

That count is of known, sequenced versions rather than everything that exists, and it will be out of date within months.

What counts as a match.

In the United States the working standard for an unrelated donor is 8/8: HLA-A, -B, -C and -DRB1, two copies of each, one inherited from each biological parent. European centers usually add -DQB1 to make 10/10, and adding -DPB1 gives 12/12 — the level at which a fully matched sibling matches.

These are conventions about how many places to compare, not competing definitions of a match. A transplant team weighs HLA alongside the patient’s disease, age, health, and which donors are actually available.

A closer match is a better transplant. In an NMDP analysis of 3,857 unrelated-donor marrow transplants performed in the United States between 1988 and 2003, one-year survival was 43% where the donor differed at one of the eight markers, against 52% for an 8/8 match. Outcomes have improved a great deal since that era; the finding that has lasted is that every mismatch costs something.

Cell-surface marker

HLA proteins are found on most cells and form part of the body's biological identification system.

Peptide-binding groove

The groove displays small protein fragments that immune cells can inspect.

Immune recognition

Immune cells use HLA-related signals to help distinguish the body's cells from cells that appear foreign.

Inherited code

A person inherits half of their HLA information from each biological parent.

The first search begins at home.

HLA genes sit close together on one chromosome, so they usually travel as a block — a haplotype — passed down whole rather than reshuffled. You have two: one from each biological parent.

That is why a half match is a coherent inherited half-set rather than a scattering of coincidences, and why a parent is almost never a full match for their own child. By definition, each parent gives only one of the two halves.

Usually, not always — the block can be broken by recombination. That is why a transplant team types every family member directly instead of inferring anyone from anyone else.

Every full sibling is a one-in-four chance.

Each parent carries two haplotypes and passes one down, so there are four equally likely combinations a child can inherit. Two full siblings match fully only if they happened to draw the same pair.

And it is why about 70% of patients who need a transplant do not have a fully matched donor in their immediate family — NMDP’s figure for the United States. Smaller families have made unrelated donors more important, not less.

Then the family is tested, and waits.

Siblings are swabbed or have blood drawn, and the result carries more than genetics. Anthony Nolan warns families outright that a relative may feel pressure to donate, and may feel guilty if recovery goes badly.

A relative who matches then donates the same way a stranger would — injections and several hours on a cell-separating machine, or a marrow collection under general anesthetic. Mark, who donated twice to his sister, told Anthony Nolan: “When you donate to a stranger, you can feel good but walk away. When it’s your family member, it’s much more personal and harder emotionally.”

And there is the particular silence of the sibling who is not a match.

Then the search becomes global.

A transplant team queries a worldwide pool. On 27 July 2026 the World Marrow Donor Association listed about 43.9 million records — registered adult donors and stored cord blood units, counted together — from 120 sources across 56 countries.

A donor search is not a national search. It is a planetary one — and it still does not find a full match for everyone.

That number deserves three cautions. It is a live counter, so it was already out of date when you read it. It adds two different things together: people who volunteered, and frozen units of donated cord blood — a unit is not a person. And a source is not the same as a national registry; some countries list several, some listings are cord blood banks, and a country absent from that list has not been shown to have no route at all. The Association itself is not somewhere you can join. It sets standards and runs the machinery that lets registries search each other.

And it does not work equally well for everyone.

Because HLA is inherited, a patient is most likely to match someone with a similar ancestral background. HLA types differ between populations because the immune system evolved alongside local disease history and human migration.

People of mixed heritage face a compounding version of the same problem: each inherited half may be common on its own while the combination is rare, or effectively unique, on any register.

This is a gap in who is represented on registries, combined with how HLA varies between populations. It is not a gap in generosity, and no group’s biology is harder to match in itself.

Recruitment alone will not close it.

NMDP modeled the best possible case in the United States: every eligible Black or African American potential donor aged 18 to 35 — 11.3 million people — joining the NMDP registry and being available when called. The 8/8 match rate would rise from 29% to 63.7%.

Still short of the 79% that non-Hispanic White patients see. The reason is population genetics: HLA diversity is greatest in populations of African descent, so many more distinct combinations have to be covered to reach the same match rate.

Which is exactly why registries pursue both things at once — recruiting more people, and advancing the science that makes mismatched transplants work. Joining is necessary. It is simply not sufficient on its own.

And being listed is not the same as being available.

NMDP reported in 2024 that about half of potential donors in the United States are unable to donate when they are matched with a patient. Fear of losing work is among the leading reasons — NMDP puts the commitment at an average of 20 to 30 hours spread over four to six weeks, not counting travel.

What stands between a listed match and a real one

• Contact details go stale. Someone who joined at a university drive at 19 can be unreachable at 26 — matched, willing, and invisible.
• Some people are deferred for medical reasons entirely outside their control.
• Some say no, which is their right, at any point.
• And a search does not stop at one name. More than one possible donor is often checked for the same patient, precisely because any one of them may turn out to be unreachable, unwell, or unwilling.

Whether those hours are protected at work depends entirely on where you live. We found two protections written into law: in the United States, federal executive-branch employees are entitled to up to seven days of paid leave a year for marrow donation, and Italian law preserves a registered donor’s pay for tests, admission, donation and recovery. We could not verify a general protection anywhere else, which is not the same as proving none exists — but it does mean you should assume there is none until you have checked.

And not everyone who wants to be on it can be.

Every registry writes its own rules, and it applies them twice — once when you sign up, and again in far more detail if you are ever matched with a patient. Those are two different tests. Clearing the first does not commit you to the second, and a rule that would stop a donation does not always stop a registration.

The thresholds are further apart than people assume. NMDP in the United States takes new members aged 18 to 35 and keeps them listed until 61. Anthony Nolan takes 16 to 30. DKMS UK takes 16 to 55.

What the rules actually look like up close

• Body size is usually judged at the point of a match, and not by one number — NMDP publishes a maximum weight for each height, while Anthony Nolan applies a BMI ceiling of 40 for a bloodstream donation and 35 for a marrow one.
• Pregnancy is nearly always a pause rather than a bar. Several registries let you join while pregnant and hold you out of searches until around six months after the birth.
• Autoimmune conditions turn on which condition. More than one registry accepts a stable, well-controlled thyroid condition while excluding active systemic disease.
• A cancer history usually excludes — with named exceptions that recur across registries, including basal cell skin cancer and carcinoma in situ of the cervix.
• Medication is rarely the deciding factor by itself. NMDP puts it plainly: it is usually not the drug, but the condition the drug is for.

Examples read from those registries’ own pages on 27 July 2026, and they change. This page cannot tell you whether you are eligible — only the registry serving your country can, and deciding is its job rather than yours. People rule themselves out far more often than the rules do.

A full match is valuable. It is not always the only path.

Transplant teams may consider full, partial or haploidentical donors, depending on the patient, the disease and what is available.

And the cells themselves can come from three places.

They can be drawn from a donor’s circulating blood after several days of injections. They can be taken from the marrow of the pelvis under anaesthetic. Or they can be thawed from a unit of cord blood, donated after a baby was born and frozen since.

Which one a team asks for is a clinical decision rather than a preference. It turns on the patient, the disease, how urgent it is, and what can actually be obtained in time — which is why the best available donor is not always the one with the highest number of matching markers.

A donor is asked which methods they are willing to do, and a registry does not proceed with one they have not agreed to.

The third source is usually thrown away.

After a baby is born and the cord is cut, blood is left behind in the placenta and the cord itself. It is rich in blood-forming cells. A cord blood unit is that blood — collected, tested and frozen. No blood is taken from the baby, and collection happens only after mother and baby have been safely delivered.

Its immune cells are less mature than an adult donor’s, and that turns out to matter. The FDA puts it carefully: there is evidence cord blood cells may not require as exact a match as cells from marrow or the bloodstream, because the antigens in cord blood are less mature. For a patient whose inherited combination is rare on the registries, that tolerance can be the difference between having an option and having none.

Greater tolerance is not the absence of matching. Compatibility, cell dose, the quality of the unit, the disease and the centre’s own experience all still bear on whether a particular unit can be used at all.

What it does not solve.

A cord blood unit cannot be asked for more. It holds however many cells it held on the day it was collected, and for a larger patient there may simply not be enough in one.

Where that runs out

• Too few cells can mean slower recovery of the blood counts and a greater chance the graft never takes hold.
• Some teams have used two units when one was not enough — though a US trial in 224 patients aged 1 to 21 found no one-year survival advantage for two, and its single-unit recipients recovered platelets faster with less severe GvHD.
• Public collection runs only at participating hospitals — four in Canada, three in England. Most maternity units in the world cannot collect at all, so for most families the answer is decided by geography before anything else.

Donating to a public bank costs a family nothing, and the unit then belongs to whoever needs it — it cannot be held back for your own family. Private banking stores it for you and charges for that. In March 2019 the American College of Obstetricians and Gynecologists said routine private banking is not supported by the available evidence and recommended public donation, treating storage for a relative with an already diagnosed need as a separate case. That is US professional guidance, not a worldwide rule. Worth knowing either way: a child’s own stored unit is not automatically the right graft for that child — the FDA notes that for genetically heritable diseases it may not be recommended, and that a bank being registered with the FDA is not the same as being endorsed by it.

What a mismatch costs.

Every marker that differs gives the arriving immune system one more thing to react to. Mismatch raises the risk of graft-versus-host disease, and the risk that the graft never takes hold at all.

Not every marker behaves the same way. In the 2007 NMDP analysis, single differences at HLA-B or -C appeared better tolerated than differences at HLA-A or -DRB1. Which places matter, and how much, is still actively researched.

Better matching is one of the main ways a transplant team reduces those risks.

And the science is moving underneath all of it.

In the NMDP-sponsored ACCESS trial, published in 2025, 145 adults at 21 US centers received cells from unrelated donors who were not full matches — most at 7 of 8 markers or 6 of 8, a few lower — using post-transplant cyclophosphamide to hold graft-versus-host disease in check.

One-year overall survival was 83.8% with myeloablative conditioning and 78.6% with reduced-intensity conditioning — comparable to historical benchmarks for fully matched transplants. Fifty-nine per cent of participants were of ethnically diverse ancestry, against under 20% in US transplant trials historically. NMDP now reports that in the United States, for adults with the common blood cancers, a suitable donor or cord blood unit can be identified for very nearly every patient — though being identified is still not the same as being available when the call comes, and access differs sharply between countries.

This does not make matching unimportant. A better match is still a better transplant, and these are trial results in one group of patients. What it changes is the meaning of “no match found” — that sentence is no longer the end of the search.

Full match

Donor and patient match at every marker compared.

Partial match

One marker differs. Transplant teams still use partial matches, often with extra steps to manage the immune mismatch.

Haploidentical match

The donor shares exactly one inherited half-set — usually a parent, child or half-matched sibling.