Can You Get DNA From Ashes? The Definitions That Decide It

Almost never. A crematory returns calcined bone that has been held between 1,400 and 1,600 °F (760 to 871 °C) for thirty minutes to two hours and then pulverized to a uniform grain, which is the sequence the Cremation Association of North America publishes on its own site. Nuclear DNA does not survive that. In an experiment that heated bone in 100 °C steps and tested for DNA at each one, reported by Harbeck and colleagues in Forensic Science International in 2011, short mitochondrial fragments survived at 600 and 700 °C and nothing usable survived above that. Commercial cremation spends its entire cycle above that ceiling, so a laboratory attempt on ashes succeeds only when something in the container escaped full exposure.

I spent years driving test routes with a logging rig in the boot, comparing what an operator's coverage map claimed against what the meter read at the kerb. The map was never lying. It answered a narrower question than the one I was asking: signal above a chosen level, outdoors, for a chosen percentage of locations. Every coverage argument I ever had ended when someone read that definition aloud. This question behaves the same way, and three words carry the load: ashes, DNA, and identification.

What is in the urn, by mass and material

An adult cremation returns roughly 900 to 3,800 grams. Warren and Maples measured 91 adult cremations in Florida for the Journal of Forensic Sciences in 1997 and reported a mean of 2,430 g, a range of 876 to 3,784 g, and a figure of about 3.5% of body weight. Bass and Jantz repeated the work in East Tennessee in 2004 across 306 cases and found heavier remains, 3,380 g for men and 2,350 g for women. Van Deest's 756 northern California cases gave 3,233 g and 2,238 g in 2011.

None of that mass is tissue. It is the mineral fraction of bone, hydroxyapatite, with the collagen burned out and the fragments run through a processor; CANA places pulverization after cooling, as a distinct step. One object in the container is not bone: the identification disc, commonly 304 stainless because its melting point clears the chamber temperature. That disc is the only thing in the urn still carrying information about who this is.

The temperature at which DNA stops being recoverable

Degradation happens in stages, and the stages have been measured.

• Above about 200 °C, the carbon-to-nitrogen molar ratio in bone leaves the 2.9 to 3.6 window that indicates intact collagen. Harbeck's team recorded that transition.
• Above about 300 °C, STR amplification falls off sharply. Garriga and colleagues reported the drop in 2016, and Lozano-Peral's group confirmed in 2021 that longer fragments degrade first.
• Between 300 and 600 °C, bone loses the bulk of its remaining mass to combustion.
• At 600 and 700 °C, short mitochondrial fragments were still amplifiable in Harbeck's experiment, with duration of exposure mattering as much as peak temperature.
• Above 1,000 °C, von Wurmb-Schwark and colleagues found in 2004 that profiling from fully cremated remains is often unreliable.

Two caveats belong with those numbers. Harbeck's team used defleshed cattle tibiae rather than human bone in a coffin, so the thresholds are indicative. Imaizumi's 2014 work on metacarpals is far more pessimistic still, getting mitochondrial amplification only up to 200 °C at a fifteen-minute exposure. The honest reading is a band rather than a line, and commercial cremation sits above the whole band.

Nuclear DNA and mitochondrial DNA answer different questions

The word "DNA" on a laboratory quote can mean either of two molecules with different survival odds and very different evidentiary reach.

A nuclear STR profile is what identification usually means. Since 1 January 2017 the FBI's CODIS core has been 20 loci, expanded from 13 by adding D1S1656, D2S441, D2S1338, D10S1248, D12S391, D19S433 and D22S1045. Twenty loci separate one person from everyone else and support a parentage calculation. Nuclear DNA is also the fragile molecule, present at two copies per cell, and the first thing heat destroys.

Mitochondrial DNA sits at hundreds to thousands of copies per cell inside a second membrane, which is why it outlasts nuclear DNA under heat. It descends only through the maternal line, identical in a mother, all of her children, her siblings by the same mother, and her sisters' children. A mitochondrial match places a sample in a maternal lineage. It cannot say which member, and it carries nothing about a father. A laboratory that recovers mitochondrial sequence from ashes has not identified anyone; it has narrowed the field to a family branch.

Cremated remains, burned bone, and a stored sample are three different specimens

Search results for this question mix hair, bone and ashes as if they were interchangeable. They are different materials and they do not produce the same result.

| Specimen | Heat exposure | What is left | Usual laboratory outcome | |---|---|---|---| | Cremated remains from a crematory | 760 to 871 °C, 30 min to 2 h, then pulverized | Bone mineral as uniform grain, plus the ID disc | No nuclear profile; mtDNA rarely, unreliably | | Calcined bone from a fire (blue-grey-white) | Above roughly 700 °C, fragment intact | Bone mineral, structure preserved | Typing succeeds only sporadically | | Charred bone from a fire (black burnt) | Roughly 300 to 600 °C | Heavily degraded nuclear DNA | Incomplete but authentic STR profiles | | Unburned bone or tooth | Ambient | Nuclear DNA at low yield | Full or near-full profile | | Reference sample taken before cremation | Ambient | Intact nuclear DNA | Routine full profile, first attempt |

Those middle rows come from Schwark and colleagues, who extracted DNA from 71 bone fragments belonging to 13 individuals and sorted the results by burn stage in Forensic Science International: Genetics in 2011. Well preserved and semi-burnt bone gave reliable identifications. Black burnt bone gave incomplete but genuine profiles in most cases. Blue-grey bone worked only sporadically. Blue-grey-white calcined bone barely produced reliable results at all, and heavily burnt fragments proved prone to contamination by external DNA.

The gap between the first row and the two beneath it explains why disaster identification statistics do not transfer. A systematic review in Genes pooling 37 studies published since 1995 reports identification rates above 90% in mass fatality work, including 96 of 97 victims in one aviation disaster. Those victims were burned. They were not cremated for two hours and then milled.

Yield figures make the same point from the other side. Laboratories typically decalcify about half a gram of bone powder per extraction, and even sound skeletal material gives very little: femurs from Second World War mass graves averaged between 0.3 and 3.2 nanograms of DNA per gram and still produced STR profiles. A two-minute buccal swab from a living relative delivers thousands of times that.

What a cremated-remains DNA test actually buys

No accredited laboratory publishes a validated success rate for cremated remains. The figures circulating online, "93% of cases" and "less than 1% success," sit on consumer comparison pages with no method attached, and I would not repeat either as a finding. What is published is Schwark's category table, and it says calcined bone barely yields reliable results.

The first fee generally buys an extraction attempt and a presence-or-absence screen, nothing more. EasyDNA and The Carlson Company both describe the comparison against a relative's profile as a separate charge. PTC Laboratories publishes a $100 non-refundable administrative fee on every order, legal or not. Cremains work carries no public price; it is quoted case by case in the low thousands, against $300 to $600 for a legal chain-of-custody kinship test.

The word that decides your invoice is "result." Before paying, get three things in writing: whether the fee covers a sequence or only a screen, what will be reported if nothing amplifies, and whether the relative comparison is included. Ask about ISO/IEC 17025 accreditation, then ask whether the accredited scope covers skeletal material, because general accreditation does not imply it.

The kinship test runs on the living, and it works

A relationship test between living people answers the question families are actually asking, with a number. AABB's Standards for Relationship Testing Laboratories require a combined paternity index of at least 100 to 1 before a laboratory may report support for a claimed parentage, which converts to a 99.0% probability under the standard formula PI/(PI+1). Real inclusion reports usually land at 99.99% or above. The US Department of State sets a higher floor of 99.5% for immigration cases.

Relatives are not interchangeable. A parent and child share exactly one allele per locus by descent, which is why the calculation is close to decisive: in one published comparison, a 15-locus STR panel separated parent-child pairs from unrelated pairs 86.75% of the time and full siblings 81.73% of the time. Second-degree relationships collapse. Using 52 sequence-typed STR loci and a threshold of LR above 10,000, a recent study distinguished full siblings from unrelated pairs 99.85% of the time, and half-siblings, grandparent-grandchild pairs and uncle-nephew pairs about 61% of the time. A grandparent and grandchild share no allele by descent at half their loci, which is the arithmetic behind that 61%.

What to do when the ashes are all you have

1. Request the chain-of-custody record and match the number stamped on the identification disc against the cremation log. CANA's described procedure checks identity at storage, loading, removal, processing, urn filling and release, so the paper trail exists to ask for.
2. Photograph the contents before anything is sifted or divided, and look for material that is not uniform grain. A tooth fragment, or cortical bone larger than the surrounding powder, is the only thing worth submitting.
3. Ask the laboratory in writing what constitutes a reportable result and what the report will say if nothing amplifies.
4. Test living relatives in parallel rather than afterwards. A first-degree kinship result stands alone, costs an order of magnitude less, and depends on nothing having survived the retort.
5. If a cremation has not happened yet, ask the funeral home to collect and retain a reference sample: blood on a card, buccal swabs, or nail clippings, air-dried and stored in paper rather than sealed plastic.

Questions families ask

How can someone tell whether cremation ashes are genuine?

Match the stainless identification disc in the container against the crematory's cremation log and chain-of-custody paperwork. The disc survives the chamber and travels with the remains from intake to release. Weight is a second check, since adult cremations return roughly 900 to 3,800 grams. DNA testing cannot settle the question.

What happens to teeth during cremation?

Permanent teeth show no microscopic change below 200 °C. Between roughly 300 and 500 °C, enamel and cementum fissure and fragment along the cemento-enamel junction and cusp margins, while dentine holds until about 700 °C. Above 800 °C the crown turns chalky white as hydroxyapatite crystals grow. Fillings, crowns and implants survive.

Can a deceased person's hair provide DNA?

A hair with root tissue can give a nuclear profile, though one laboratory reported success in only 32% of rooted hairs, rising to 69% after adding a hematoxylin staining screen. Rootless shafts give mitochondrial sequence reliably: one study recovered over 98% of the mitogenome from more than 94% of samples, including 27-year-old hairs.

Can DNA be recovered from bone?

Yes, routinely, from unburned bone. Laboratories decalcify about half a gram of bone powder and work with tiny quantities: femurs from Second World War mass graves averaged 0.3 to 3.2 nanograms of DNA per gram and still yielded STR profiles. The petrous portion of the temporal bone, metacarpals and metatarsals give the highest yields.

Can cremated remains identify a person?

Not through DNA, in ordinary cases. Cremated remains are pulverized calcined bone whose nuclear DNA has been destroyed by two hours above 760 °C, and published work on calcined bone reports only sporadic typing success. Identification rests on the crematory's identification disc and chain-of-custody records, or on testing living relatives.

Which living relative is most informative for a kinship test?

A child or a parent of the deceased. First-degree relatives share exactly one allele per locus by descent, which standard STR panels resolve reliably. A full sibling ranks next. Grandparent-grandchild, half-sibling and uncle-nephew pairs were separated from unrelated pairs only about 61% of the time in recent testing.