Every few months I catch a true crime series on a late-night scroll, and the same scene repeats: the lab, the glowing gel, the match. The suspect walks. The killer walks. It feels like magic — and honestly, for the first decade of DNA in forensics, it kind of was. But the real story isn’t the TV version. It’s a geneticist’s accident, a man who confessed to a murder he didn’t commit, and a serial killer who stayed free for decades because nobody thought to check a genealogy database.
Here’s how it actually works, and where it gets surprising.
It started with an accident, not a plan
In 1984, British geneticist Sir Alec Jeffreys was running research at the University of Leicester when he noticed something odd in his samples: regions of DNA containing repeated sequences — minisatellites — that varied wildly from one person to the next. He realized the number and arrangement of those repeats could be used to identify a person, the same way a fingerprint identifies a palm. His team built radioactive probes that bound to those repetitive regions and revealed a pattern so distinctive it was unique to each individual. Jeffreys called it a DNA fingerprint. He also noticed children inherit half of their pattern from each parent — which meant the technique could prove family relationships, not just identity.
The first real-world test wasn’t a crime at all. In 1985, a Ghanaian family in Britain had a son who’d traveled to Ghana and was denied re-entry because immigration officials doubted he was really their child. Jeffreys analyzed DNA from the mother, the boy, and several siblings. The patterns proved he was her biological son. It was the first practical use of DNA fingerprinting — and it cleared an innocent man’s passport problem before the tech ever met a crime scene.
The case that proved DNA both ways
Two teenage girls, Lynda Mann and Dawn Ashworth, were raped and murdered in Leicestershire — in 1983 and 1986. Because the crimes were so similar, investigators believed one man did both. A local teenager, Richard Buckland, confessed to the second murder. He seemed like the answer, so police asked Jeffreys to analyze the biological evidence from both scenes.
He was nervous. His original method needed relatively large amounts of reasonably intact DNA, and crime-scene samples are rarely that generous. But his team ran it anyway, and the results went two ways at once. The DNA proved the same man committed both murders — and Buckland’s DNA didn’t match either scene. He was released, the first person ever exonerated from a major case by DNA evidence. The actual killer was still out there.
So police ran one of the first large-scale DNA screenings in history. Thousands of local men were asked to volunteer blood or saliva samples. Approximately 5,000 men were tested. No match. Then a man named Ian Kelly was overheard telling a friend he’d given a sample on behalf of someone else: Colin Pitchfork. Police took Pitchfork’s DNA, compared it to the crime-scene evidence — match. He confessed. In 1988 he became the first person convicted of murder largely on the basis of DNA fingerprinting.
One case, and the technology had just shown its full range: it can save the innocent and catch the guilty. Both in the same file.
The upgrade: from big samples to tiny ones
Through the late 1980s and 1990s, two advances made DNA profiling practical on a national scale. The first was the polymerase chain reaction — PCR — which lets scientists make millions of copies of a small DNA segment, so labs could work with samples far smaller than Jeffreys’s original method allowed. The second was short tandem repeat analysis, or STR profiling: examining short repeating sequences at specific spots in the genome. Because modern labs can test many STR locations at once, the odds of two unrelated people sharing the same profile are extraordinarily small. Modern techniques can also pull usable profiles from much smaller, sometimes degraded samples.
Cost dropped, speed went up, and governments started building databases. In the US, the DNA Identification Act of 1994 authorized the FBI to establish a national DNA index. The FBI’s Combined DNA Index System — CODIS — lets federal, state, and local labs store and compare profiles. Its national level, the National DNA Index System (NDIS), went operational in 1998. It holds profiles from convicted offenders, certain arrestees and detainees, forensic evidence from crime scenes, missing persons, unidentified remains, and relatives of missing persons. By the mid-2020s it contained tens of millions of profiles — one of the largest forensic DNA databases in the world. A match can identify a suspect or connect crimes across state lines, even decades apart.
The cold case that genealogy solved
Then came the case that changed everything about where you look for a match.
Between the 1970s and 1980s, a serial offender committed a long string of rapes, murders, and burglaries across California. Biological evidence was recovered from several scenes. For decades, that DNA matched nobody in the traditional law-enforcement databases. The case went cold — until investigators tried something new: investigative genetic genealogy.
They uploaded genetic information from the crime-scene DNA to GEDmatch, a genealogy database that, at the time, let users compare their DNA with other participants. It didn’t find the killer. It found his distant relatives. Genealogists and investigators then used public records to build family trees connecting those relatives back to common ancestors, narrowing the list of possible suspects by age, sex, location, and family structure. The focus landed on Joseph James DeAngelo, a former police officer who had lived in several of the areas where the crimes happened. Police put him under surveillance and secretly collected items he’d discarded. DNA from those items matched the original crime-scene evidence. DeAngelo was arrested in 2018. He pleaded guilty to 13 murders and admitted responsibility for numerous other crimes; investigators have connected the crime series to more than 50 sexual assaults.
A killer who beat a national database for decades. Caught through a family tree.
When the technology gets it wrong
DNA profiling is extremely reliable when done correctly. But it has failed, and it’s worth knowing how. The most famous example is the Phantom of Heilbronn. Starting in the 1990s, DNA from an unidentified woman showed up at dozens of crime scenes across Germany, Austria, and France. Investigators believed one woman was connected to a string of crimes, including murders. After years of investigation, the mystery DNA turned out to belong to a woman who worked at a factory that manufactured the cotton swabs police used to collect evidence. The swabs were contaminated during production. A brilliant technique, undermined by a manufacturing flaw.
Mistakes also happen with mixed samples — when a single piece of evidence contains DNA from multiple people. Those mixtures are hard to interpret, and analysts sometimes have to make statistical calls about whether a person’s DNA fits the mixture. And privacy is a real issue: DNA carries far more information than a fingerprint. It can reveal biological relationships, ancestry, and some inherited traits. Large government databases raise questions about how genetic information is collected, stored, searched, and retained — and familial searching can put people under suspicion they never consented to, just because a relative is in a database.
FAQ
How accurate is DNA profiling? Extremely reliable when collected, handled, and interpreted correctly. The most common failure modes are contamination (like the Heilbronn swabs) and difficulty interpreting mixed samples from multiple people.
What’s the difference between CODIS and GEDmatch? CODIS is the FBI’s law-enforcement database of profiles from offenders, arrestees, and crime scenes. GEDmatch is a consumer genealogy database. The Golden State Killer was solved by matching crime-scene DNA against GEDmatch relatives — a step traditional databases never took.
Can DNA clear an innocent suspect? Yes — and it has, starting with Richard Buckland in 1988. Exoneration is one of DNA’s quietest and most important jobs.
Does the Golden State Killer case mean all cold cases can be solved this way? Not automatically. It works when a suspect’s relatives have uploaded DNA to a consumer genealogy service, and public records are rich enough to build the family tree. But it turned “unsolvable” into “not solved yet” for a huge category of cases.
Which cold case do you think DNA should reopen next? Let me know — I read everything, and I have strong opinions about which true crime series are worth your weekend.
a note from the editor — I’ve spent eight years cutting through “miracle product” claims on this blog. DNA forensics is the rare miracle that survived contact with the lab. It’s cleared the innocent, caught the guilty, and still surprises me. Less shelf, more glow.