Her Killer Lived Next Door for 33 Years. Until DNA Exposed Him
Her Killer Lived Next Door for 33 Years. Until DNA Exposed Him
A 78-year-old woman was raped, strangled, and murdered in her own home. Then she was hidden, shoved under her own bed, as if that would buy whoever did it more time. Police canvassed the neighborhood. They talked to everyone who knew her, everyone who might have seen something that night.
Nobody had answers, no suspect, no motive that made sense. The case went cold, and it stayed that way for over three decades. For 33 years, Mary Kelly’s family waited for a phone call that never came. One that would finally tell them who did this and why. When that call finally came, the answer had been in their faces the entire time.
Her name was Mary Hague Kelly. She was 78 years old, and she lived alone in a quiet, close-knit part of Oak Cliff, a neighborhood in Dallas, Texas. Her husband had died years earlier of a heart attack, and she’d been on her own since. Before that, she made her living as a seamstress, sewing drapes for interior designers around the city.
People who knew her used the same words over and over. Warm, full of energy, not just a neighbor. The whole block considered her family. Her own family took turns checking in on her, making sure she was all right, because she lived by herself, but she was never really left alone. On the evening of January 19th, 1989, one of those routine check-ins didn’t go the way it always had.
Mary hadn’t answered her phone all day, and it wasn’t like her. Her family noticed. Her sister sent her husband, Mary’s brother-in-law, over to check on things in person. When he pulled into the street, something was already wrong. Mary’s car wasn’t parked in the driveway. It was always there. He had a key to the house, so he let himself in and called out for her.
Nothing answered back. Inside, everything looked ordinary at first. No broken windows, no signs anyone had forced their way in, nothing knocked over or disturbed. But the dogs wouldn’t settle. They kept pacing back toward the bedroom, agitated in a way that felt off. He followed them in. Under one of the guest beds, there was a suitcase Mary normally kept stored away, out of sight.
It was sticking out just slightly further than it should have been. He knelt down to look. That’s where he found her. Mary Hague Kelly had been hidden underneath her own bed. When police and the medical examiner’s investigators arrived, they had to lift the bed frame to pull her out. There was pooling of blood beneath her, enough to tell them she’d been there for hours, at least.
She had defensive wounds on her hands and clear ligature marks around her neck. Investigators knew right away this wasn’t a natural death. It was a crime scene. The family learned what had happened to her in stages. First, that she was gone. Later that same night, the rest of it. What had been done to her before she died.
For the people who loved her, the second piece of news landed harder than the first. At her funeral, family members remembered her hands being covered out of respect because of the damage the attack had left on them. She was buried in a crypt alongside her husband, the two of them together again after years apart.
Afterward, for weeks, someone in the family called the detective assigned to the case checking in. The answer was always the same. Nothing yet. You’ll be the first to know if that changes. In the beginning, that kind of waiting still felt bearable because surely someone reasoned the person responsible would be found and arrested and made to answer for it.
That certainty didn’t last. The autopsy confirmed what the scene had already suggested. Mary had been strangled. The medical examiner found fractures in her thyroid cartilage and her hyoid bone, the small bone in the neck that supports the base of the tongue. She also had petechiae, tiny burst blood vessels around her eyes, a common finding in cases of manual strangulation.
She had also been sexually assaulted. A sexual assault kit was collected during the autopsy and lab testing on her robe came back positive for semen. In 1989, DNA testing as we know it today barely existed. Investigators could confirm biological evidence was there. They had no way yet to turn it into a name.
There were no signs of forced entry anywhere in the house. No broken locks, no shattered glass. Whoever came into Mary’s home that night either got in through a door she believed was safe or she let them in herself. Some of her belongings were missing. Her purse, her jewelry, an antique phone she kept in the house, and her car, a 1980 Chevrolet Monte Carlo, had been taken as well.
The car turned up a few days later abandoned in a church parking lot not far from Mary’s neighborhood. Its windows had been broken out and the radio had been stripped from the dashboard. Inside, investigators found empty beer cans, items that didn’t belong to Mary and had no obvious connection to her at all.
The car was processed for fingerprints and any other evidence that might open up a lead. Nothing usable came out of it. In the bedroom, near where she was found, investigators recovered cords that could have been used in the attack. When those cords were eventually tested for DNA, they carried something investigators didn’t expect.
Not one unidentified male profile, but two. On two separate cords. That single detail complicated everything. It meant investigators weren’t just chasing one unknown man. They had to consider whether two people had been involved, or whether one man’s DNA had somehow ended up on both. And there was no way with the science available at the time to answer that question with any certainty.
Somewhere in that evidence was very likely the answer to who did this. It was sitting in a police lab in 1989 in a form nobody yet had the tools to read. The canvassing that followed turned up nothing solid. But it wasn’t the only thing investigators tried. They lifted fingerprints from throughout the house and submitted them for comparison.
None matched anyone with a criminal record on file. They looked closely at everyone connected to Mary, including briefly her own brother-in-law, the man who had found her. He was cleared quickly, but for a family already in shock, even that brief suspicion added to the weight of what they were carrying. The case was submitted to the FBI’s violent criminal apprehension program, a national system meant to flag similar crimes in case this wasn’t an isolated attack.
There had been talk around that time of other elderly women in the Dallas area being targeted in similar ways. Home invasions, sexual assaults, and murders with a comparable pattern stretching across the 1980s and into the following decade. Some investigators who worked Mary’s case believed at the time that it might be connected to those other crimes.
One person responsible for several attacks rather than one isolated act. That theory was never confirmed. Officers also visited pawn shops across the city hoping someone had tried to sell off her stolen jewelry. Nothing ever turned up. For the people who lived on that block, an uneasy feeling settled in and didn’t leave.
Someone had gotten close enough to walk into Mary’s home without raising any alarm, whether that meant a stranger who had studied her routine or someone who already belonged there. Neighbors started looking at each other differently, wondering if they’d ever cross paths with him without knowing it. Nobody had an answer.
And for a long time, nobody would. Weeks passed, then months. Every lead the police followed led nowhere. By the spring of 1989, the investigation had worked through nearly everything it could. There were no more names left to check, no more doors left to knock on. The case went cold. But the evidence itself didn’t disappear.
It stayed exactly where it was, waiting on science that in 1989 hadn’t caught up yet, and that no one could promise ever would. Years passed. Mary’s family held on to hope that never fully went away, even as the silence stretched on and on. Whenever they called to check in, the answer barely changed from one year to the next. Still nothing. Still no name.
For a long time, that’s exactly how it stayed. Anniversaries came and went, year after year. The case no closer to an answer than it had been the week Mary died. In 2004, the case was reopened. The medical examiner who had handled Mary’s original autopsy, Dr. Jeffrey Barnard, still had records from 1989, and he noticed something important.
Vaginal swabs had been collected at the time. Mary’s robe had tested positive for semen. DNA testing, which had barely existed back then, had come far enough by 2004 that a usable profile might finally be possible. It worked. Investigators developed a male DNA profile from the old evidence and uploaded it into CODIS, the national database used to match crime scene DNA against convicted offenders and other unsolved cases.
Then they waited for a match. It never came. Whoever had killed Mary Kelly had never been arrested for anything that would have put his DNA into that system. His profile existed now, specific and unmistakably his, and there was still no name attached to it. For investigators, this was harder to sit with than the silence of the first investigation had been.
They finally had a real scientific fingerprint of the man responsible, and there was still no way in 2004 to turn that fingerprint into an actual person. The case went quiet again. Not forgotten this time. Just waiting on a break in the science, something that hadn’t caught up yet to what this case needed.
Dr. Barnard didn’t let it go even after 2004 came up empty. He believed there were other cases like Mary’s sitting in old police files across the city, cases from the same period with a similar pattern that might eventually be solved the same way. He went back through years of manual case logs by hand and built a spreadsheet of unsolved sexual assault homicides, cases he thought could benefit from forensic testing once the technology existed to make use of it.
Mary’s case was on that list. In November of 2021, the Dallas County District Attorney’s Office tried something new. A team within the office dedicated to unresolved sexual assault cases applied for a grant from a nonprofit called Season of Justice hoping to fund DNA analysis that hadn’t existed in 2004, let alone 1989.
The grant was approved in early 2022. Mary’s case was sent to Astrium, a private forensic laboratory in Texas built specifically for cases like this one. Using a method called forensic grade genome sequencing, Astrium scientists built a far more detailed genetic profile than the standard one used for matching against criminal databases like CODIS.
That extra detail is what makes genealogical tracing possible in the first place. What Astrium does is different from a standard DNA match. Instead of comparing a profile directly against a database of known offenders and hoping for an exact hit, investigative genetic genealogy works backward through family trees.
Even a small amount of degraded decades-old DNA can reveal a person’s genetic relatives, sometimes distant ones. People connected only through a shared great-grandparent or a handful of cousins several generations removed. Investigators built a genealogical profile from the crime scene evidence and searched it against genetic databases.
It surfaced a relative, someone who shared enough DNA to suggest they were roughly a second cousin of the man they were looking for. That relative had one parent with European ancestry and one with Latin American ancestry, which meant the genetic path back to the suspect most likely ran through the Latin American side of the family.
From there, the work was less like a lab test and more like tracing a family history by hand. The relative also shared DNA on the X chromosome, which people inherit only from their mothers. That single detail let genealogists eliminate a large portion of the family tree outright, narrowing the search down to one specific grandmother’s line.
The problem was the grandmother had 11 siblings, which meant 11 separate branches of descendants to work through one at a time, cross-checking each against public records. It wasn’t fast and it wasn’t simple, but piece by piece the search narrowed down to one specific family. And then investigators found something that connected the entire search back to a place they already knew.
The family they had traced led to a house. A house that had sat right next door to Mary Kelly’s back in 1989. David Rojas’ father had lived in that house at the time of the murder. David Rojas himself, only around 20 years old that year, had lived there, too. For over 30 years the answer to who killed Mary Kelly had lived a few feet from her own front door.
Investigators finally had a name. What they didn’t have yet was him. By the time investigators identified David Rojas, he was 53 years old living in Del Rio, Texas, several hours from Dallas. He had no idea he had become the focus of a decades-old murder investigation. The FBI’s Dallas division worked with local police in Del Rio to place him under surveillance.
For several days investigators watched him waiting for a chance to collect his DNA without ever alerting him to what was happening. They needed something he’d willingly thrown away, something they could legally collect and test. The opportunity came when Rojas took out his trash. Investigators watched him discard a bag and once he was gone, they retrieved it.
Inside was a beer bottle. That bottle went straight to a lab. The results came back as a complete match to the male DNA recovered from the crime scene in 1989. On July 22nd, 2022, David Rojas was arrested and charged with capital murder. He was held at the Dallas County Jail on a $750,000 bond. His attorney at the time, Sherard Edwards, declined to comment when reporters reached out.
When the news reached Mary’s family, it landed the way 33 years of waiting might land on anyone. One family member described dropping to their knees when the call came, then going inside to tell their own mother, whose face went pale at the news. It was relief and disbelief at the same time, arriving after so many years that some of them had stopped letting themselves hope for it at all.
For investigators, it was the end of a case that had outlasted the careers of some of the officers who’d worked it. For Mary’s family, it meant something they had stopped believing would ever happen. There was finally a name attached to what had happened to her. But an arrest wasn’t the end of it. It was the beginning of something else.
Rojas was now facing a capital murder trial for a crime committed more than three decades earlier, one that would have to be tried under the laws that existed back in 1989. And the case against him wasn’t built on one clean DNA match alone. There were still two different unidentified male profiles recovered from the cords found at the crime scene.
And prosecutors would have to explain that complexity clearly enough for a jury to understand exactly what it meant, and just as importantly, what it didn’t. Whether that would be enough to convict him 33 years after Mary’s death was still an open question. The trial began in November of 2024, more than 2 years after Rojas’s arrest and 35 years after Mary Kelly’s death.
It was the first case in Dallas County history built on evidence from investigative genetic genealogy, which meant there was no local precedent for how a jury might respond to it. The prosecution was led by Layton D. Anthony, the same attorney who had pushed for the genetic testing years earlier.
Dr. Jeffrey Barnard, the retired chief medical examiner who had preserved the original evidence back in 1989, took the stand to describe what he’d found decades earlier and why he’d held onto hope that science would eventually catch up to it. Lead investigator John Wakefield also testified, walking the jury through the case’s long path from the original crime scene all the way to a beer bottle in a trash bag outside a home in Del Rio.
Prosecutors weren’t seeking the death penalty. Under the law that existed in 1989, a capital murder conviction without a death sentence carried an automatic life sentence, one that made a defendant eligible for parole only after serving 40 years in prison. A sentence of life without any possibility of parole didn’t yet exist under Texas law at that time.
The jury deliberated for a little under 2 hours. They found David Rojas guilty of capital murder. Throughout the trial, he showed little visible reaction, sitting through testimony from the medical examiner, the lead investigator, and eventually the victims of his other charges without any outward response.
Judge Layla Lawrence Mayes sentenced him to life in prison with parole eligibility only after four decades behind bars. Given his age, that amounted to a life sentence in every real sense. Of the 12 jurors, one of the six men stayed behind afterward to hear the victim impact statements. He later said the DNA evidence had already convinced him of Rojas’s guilt, but what he heard that day made the decision easier to live with.
After the verdict, something else came to light. Separate from Mary’s case, Rojas was also facing charges of sexually abusing two young family members, girls who had been around 9 and 12 years old at the time. Once he was already in custody for Mary’s murder, they came forward. In court, both women addressed him directly, describing what he had done to them and to their family.
One of them told him plainly that this would be the last time he ever laid hands on a woman again. For a jury made up of strangers, none of whom had ever known Mary Kelly, the case had started as a stack of documents, a DNA profile, decades of paperwork. By the time they had delivered their verdict, it had become something else.
Justice, delivered 35 years late to a woman none of them had ever met. For Mary’s family, closure came slowly and in pieces. Her sister, Anita Stinchcomb, had spent decades calling the department, gently asking if there was anything new to report. For years, she got the same answer every time. Still nothing.
When it finally changed, it changed everything. This was the first time Dallas County secured a murder conviction built on investigative genetic genealogy. Faith Johnson De Antoni, the prosecutor who pushed for the testing back in 2021, said afterward that pursuing justice sometimes means going outside the usual process, and that doing so was the only reason a 35-year-old file finally got an ending.
The investigators who worked it have said they hope this won’t be the last case like it. Cold case files, some even older than Mary’s, are still sitting in evidence storage across the country, waiting for the same break that finally reached her. None of that changes what happened to Mary Kelly or how long it took to name the man responsible.
It only changes what her family gets to carry forward. Not a question anymore. An answer, even one that came far later than it should have. Long before any of that, before the DNA, before the trial, Mary Hague Kelly was simply a woman who sewed drapes for a living, who checked in on her neighbors, and who was checked in on in return.
She was somebody’s sister, somebody’s grandmother, someone an entire block considered family. For 35 years, the only certainty in this case was that someone had taken her life and gotten away with it. That isn’t true anymore. What remains is the plain fact of how close the answer had been the entire time. Not hiding somewhere far away, just next door.