In January 1989, police in Dallas found the body of 78-year-old Mary Hage Kelley inside her home on North Francis Street in Oak Cliff. There were no signs of a break-in. That single detail shaped the investigation for decades. If no one had forced their way in, detectives reasoned, Mary must have opened the door for someone she knew, a stranger, a repairman, someone passing through.
But the person they overlooked lived right next door. Mary was a tiny woman, barely four and a half feet tall and about 75 pounds, but family remembered her as someone who filled a room. She worked as a seamstress, making custom curtains for interior designers around Dallas, work that took decades of skill to build a name on. Relatives described her as energetic and quick to laugh, a woman who, in their words, knew no enemies.
Her husband had died suddenly of a heart attack years earlier, and afterward she lived alone in the house they had shared. Her family worried about that, so they took turns staying with her, checking on her, making sure the widow was never alone for long. For the neighborhood, she became a kind of grandmother to everyone, the small, spirited matriarch of the street, someone nearly every family had knocked on the door of at one time or another. On January 19, 1989, Mary had not answered her phone for days.
Concerned, her sister sent her husband that evening to check on her. Before he even reached the door, he noticed her car was not parked in its usual spot. He let himself in with his own key. Nothing inside looked suspicious.
He called her name. No answer. It was the family dogs that led him to the back bedroom, restless and pacing. He followed them and saw a suitcase, normally stored out of sight under a guest bed, pulled slightly into view.
He bent down and found her. When police and a forensic pathologist arrived, they determined the killer had made considerable effort to hide her. Investigators had to lift the bed frame to recover the body. The blood pooling suggested she had lain there for at least several hours.
She had abrasions and defensive wounds on her hands, and marks around her neck. Signs of strangulation were immediately clear. Investigators also recognized evidence consistent with a sexual assault. No cord was found around her neck, but several loose cords lay in the bedroom, one right beside her body.
To investigators, that detail alone suggested this was not random. Whoever did this had used something already in the house. The autopsy, performed by Dr. Jeffrey Barnard, then one of Dallas County’s forensic pathologists, confirmed what the scene suggested.
Mary had fractures in her thyroid cartilage, larynx, and hyoid bone, and petechiae, tiny blood spots in her eyes caused by sustained pressure on the blood vessels in the neck. Medically, those findings were consistent with manual strangulation or strangulation with a cord. She also had scrapes on her buttocks and bruising and injuries in the vaginal region. A complete sexual assault kit was collected.
The medical conclusion was that this was a rape that ended in death. Some of Mary’s jewelry, her handbag, and an antique telephone were missing, along with her car, a 1980 Chevrolet Monte Carlo. It was found two days later abandoned in a church parking lot, windows smashed, radio removed, and a case of beer cans in the trunk, cans her family insisted did not belong there. In the corner of her bedroom, a closet door had been pulled from its hinges, revealing an old analog clock radio.
At some point during whatever happened in that room, the radio had been yanked from the wall with such force that it had stopped completely. It was frozen at 3:23. Investigators believed that if there had been a struggle violent enough to tear the radio from the wall, that time probably marked roughly when it happened. It remains the most concrete piece of evidence in a crime with no witnesses.
Here was the first gap in the case, and it was not a broken lock or an open window. It was the year itself, 1989. Forensic testing of the cord found near Mary’s body showed male DNA. A preliminary test of her bathrobe indicated the presence of semen.
Each finding should have meant a great deal. But in early 1989, DNA profiling barely existed as an investigative tool. There was no national database to compare a genetic sample against, and no way to turn unknown biological material into a name. All investigators could do was secure the evidence and hope technology would eventually catch up to what they had collected.
Meanwhile, Dallas police canvassed the neighborhood. Some neighbors mentioned a car parked oddly in the alley behind the house. Others said Mary’s back door had been open. As with any murder investigation, detectives looked closely at the family, including the brother-in-law who had found her.
He was questioned and cleared. At the time, some Dallas detectives quietly compared this case to a series of other unsolved murders of older women living alone in the city during the 1980s and 1990s, break-ins with sexual assault that followed a similar pattern. Whether Mary’s murder was connected was never confirmed. Dr.
Barnard, who had arrived in Dallas in 1987, was one of those who suspected a link. Years later, when DNA testing became possible, he went back through old paper police files, which had no organized digital system, and built his own handwritten spreadsheet of unsolved cases, hoping the new technology might link several to the same offender. Nothing produced a name. The latent fingerprints found in the house matched no one on file.
Without the ability to compare the DNA evidence with a database, and without a witness naming a suspect, the investigation, which had run for about two to three months, went cold. Mary’s family did what families in that situation often do. They waited, and they called. For years they periodically contacted the detective assigned to the case.
The answer was always the same: nothing yet, but you will be the first to know if something comes up. In 2004, the Dallas Police Department officially reopened Mary’s case. By then, DNA science had advanced enormously. Dr.
Barnard, who said he never fully let the case go, remembered that a complete evidence kit had been preserved from 1989, the vaginal swabs, the bathrobe, everything. He pushed for the biological evidence to be tested with newer technology. It worked, to a point. Eventually, a usable male DNA profile was developed from the evidence.
That profile was uploaded to CODIS. This time, unlike 1989, there was a database to search. It contained no answer. There was no match.
The person who left that DNA had apparently never been arrested for a crime requiring a sample. The profile was real, and it pointed to no one. That was the second problem. Even strong evidence is only as useful as the data you have to compare it against.
The case went quiet again. Years later, a new field of forensic science began solving cases that CODIS alone never could: investigative genetic genealogy. Rather than comparing DNA directly to a criminal database, this method uploads a genetic profile to commercial genealogy databases and builds a family tree from anyone who appears as a distant genetic relative, a second cousin, a great-aunt, someone with no criminal record and no idea they are about to become the key to unraveling a decades-old murder. For many departments, the problem was not the science but the cost.
Dallas County had no budget for this kind of testing on a 1989 cold case. Layton D’Antoni, chief of the investigation division, and Jonathan Wakefield, senior detective with the Dallas County District Attorney’s Cold Case Division, sought outside help. Through a partnership with the medical examiner’s office, they were connected with Astrium, a forensic genetic genealogy laboratory. Astrium applied for a grant on behalf of the case from Season of Justice, a nonprofit that funds exactly this kind of testing for unsolved cases.
The application, including the necessary DNA extracts, was submitted in 2021. Early in 2022, funding was approved. The lab took a DNA extract originally collected from Mary’s bathrobe in 1989 and performed full forensic genome sequencing. It worked.
The result was a usable, high-quality profile that could be uploaded and compared to a genealogical database. The profile matched a woman who was genetically about a second cousin of the unknown man. The match led to a family tree, and one branch of it stood out. Two brothers from that family lived right next door to Mary when she was murdered.
The family name was Rojas. David Rojas had a criminal record that included a prior allegation of sexual assault and reports of child abuse. At that point, he lived about 400 miles away in Del Rio, Texas. Investigators with the FBI in Dallas and local police in Del Rio moved carefully.
They needed a DNA sample without alarming him. They watched him for two to three days, waiting for a lawful opportunity. It came in the form of his own trash, discarded beer cans he had set out at the curb. Legally abandoned property, they had the right to collect and test it.
On June 28, 2022, the DNA from those cans was a complete match to the profile developed from the 1989 crime scene evidence. On July 22, 2022, David Rojas was arrested. He was 53 years old, old enough, investigators noted, that had this case never been solved, he could still have been capable of committing new violent crimes. A cheek swab taken after his arrest was directly compared to the original DNA extract from the bathrobe.
It was a match without question. In 1989, David Rojas was about 19 or 20 years old. He had grown up next door to Mary and, according to several family members, came and went regularly from the house beside hers. He was never identified as a suspect.
Family members recalled that detectives spoke with his brother and father, who lived next door, but as far as they knew, not with David Rojas himself. No broken window this time, just a familiar face, someone close enough, young enough, and ordinary enough that in a neighborhood where everyone knew everyone, no one noticed him. When David Rojas gave a statement to Dallas police, he offered an explanation for his DNA. He claimed he and Mary had a sexual relationship for about two to three years, beginning when he was 17, 18, and 19 years old.
He said he mowed her lawn and worked on her car, that she could not pay him, and that she offered sex in exchange. That was David Rojas’s story. Follow-up investigation did not support it. Detectives returned to neighbors and David Rojas’s own family members.
None of the neighbors remembered him ever mowing Mary’s lawn. A man who lived across the street said plainly that David Rojas never did that, that other neighbors did the yard work themselves. David Rojas’s own ex-wife told investigators something even more direct: he never mowed anyone’s lawn, not even their own. She was the one who mowed their grass.
And Mary’s family said she had never had another relationship after her husband died, certainly not with a teenager decades younger than her. The claim of a consensual relationship stood as a defense built entirely on the word of one man, against a woman who could no longer speak for herself. The trial of David Rojas began in early November 2024, more than two years after his arrest and 35 years after Mary’s murder. It was reportedly the first murder trial in Dallas County history based on a case originally brought to light through genetic genealogy.
Rojas was 55 years old. Because prosecutors chose not to seek the death penalty, a conviction for murder would automatically mean life in prison without parole. During the trial, roughly twelve to fifteen witnesses testified, including many of Mary’s family members, who spoke about who she was and what she meant to the people around her. Investigators emphasized the physical evidence, particularly the direct STR DNA comparison that had first identified David Rojas.
The genetic genealogy was not presented as evidence in court. It had done its job by leading investigators to a name; the case rested on the traditional DNA match. The defense maintained the relationship was consensual. Texas law made that argument hard to fully refute, for one unavoidable reason.
The only person who could say with certainty it was not consensual was no longer alive to testify. The jury did not take long. After closing arguments around 1:00 p. m.
, a verdict was reached within a few hours. On November 7, 2024, David Rojas was convicted of murder. He received a life sentence without the possibility of parole. He was also formally charged with two counts of indecency with a child involving two female relatives, who were reportedly about 9 and 12 years old at the time of the alleged offenses.
Family members spoke during sentencing about the accusations, saying they had only felt able to speak once he was behind bars. For Mary’s family, the resolution did not erase more than three decades of uncertainty, but it finally answered a question some of them had carried their whole lives. One relative said the hardest thing they ever did was forgive him, not for his sake, but for their own, in their search for peace. The case also carried meaning beyond her family.
According to the Dallas County District Attorney’s Cold Case Division, it was the first case in the district’s history to produce a murder conviction based on investigative genetic genealogy. The case is now cited in the broader push for public funding to apply this kind of work to the thousands of unsolved cases still waiting in evidence rooms across the country. Investigators and advocates have pointed to the federal Carla Walker Act, championed by Texas Senator John Cornyn, as the sort of sustained government funding the technology still needs, rather than relying on nonprofit grants and per-case fundraising. Mary was buried in a yellow dress, in a crypt beside the husband she had survived by only a few years.
Finally reunited, her family says. It took 35 years for the justice system to acknowledge that. Mary Hage Kelley was murdered by someone who lived so close that he heard her dogs bark, whose family had lived beside hers for years, someone so completely absorbed into that close-knit neighborhood that no one ever pointed a detective toward him. Investigators cannot say with certainty how that night began.
One theory, offered as speculation and never confirmed, is that David Rojas entered through a back window, one that had been repaired unprofessionally, a board covering a crack in the glass that appeared to predate the murder. In that scenario, his original intention may have been nothing more than stealing from a house he already knew well. If so, it was the opportunity itself that turned a burglary into a murder. A widow alone at night, and a young man from next door who decided in that moment to take far more than possessions.
He exploited proximity. He exploited trust. And for more than thirty years, he was protected not so much by his own caution as by a series of technological and financial gaps in the justice system, gaps that had nothing to do with how well he hid his tracks, and everything to do with what science and funding simply could not yet solve. Those gaps have been closing, case by case, in recent years.
Mary’s case is proof of how far the technology has come, and a reminder of how many other names are still waiting in evidence vaults for the next gap to close.