On November 4, 2024, investigators in western Wisconsin received a name that had never appeared anywhere in their files: John Keith Miller, an 84-year-old man living in an assisted care facility in Owatonna, Minnesota. The name came from a genealogical research center in New Jersey, and it finally gave detectives the answer to a question they had been asking for half a century. Five decades earlier, in February 1974, a 25-year-old graduate student named Mary Schlais had been stabbed to death in a rural ditch in Dunn County, Wisconsin. For fifty years, the case had gone nowhere.

The evidence that would eventually solve it had been sitting in a Wisconsin evidence room since the very first day of the investigation. Mary Kathleen Schlais grew up in Champlin, Minnesota, with two brothers. She graduated with honors from the University of Minnesota College of Art and was working toward a master’s degree. She spoke multiple languages fluently and divided her time between painting and horses.

By early 1974, she was living alone in Minneapolis’s Uptown neighborhood, balancing her studies and her art, and looking forward to an exhibition in Chicago. Hitchhiking across state lines was common practice in the early 1970s, an ordinary way for young people to travel. Mary planned her route with Madison as her first stop before continuing toward Chicago. On the morning of February 15, 1974, she left her apartment for the last time.

Somewhere along the way out of Minneapolis, she accepted a ride from a stranger. By mid-afternoon, that ride had taken her about ninety miles from her apartment to rural Dunn County, Wisconsin, far off the road to Madison and Chicago that everyone expected her to take. At the intersection of 990th Street and 408th Avenue in the town of Springbrook, a neighbor named Denny Anderson was driving down the quiet road when he came upon a parked vehicle. A man stood near the car.

Mary was on the ground nearby. Anderson did not stop. As he passed, the man looked directly at him. Their eyes met for only a moment, but it was enough for Anderson to memorize the stranger’s face and the shape of the vehicle, and to understand that what he had just seen had to be reported immediately.

By the time Anderson reached a phone, the man and his car were gone. When Dunn County deputies arrived, they found a murder scene. Mary had died from stab wounds. No weapon was found at the site.

Her coat and bag were missing and were never recovered. Near where Mary was found, investigators discovered an additional object: a wool stocking cap left in the snow. In 1974, no forensic technique existed that could extract the information that cap would later yield. It was logged as evidence and stored as standard procedure, not because anyone believed it might solve the case that day.

What the investigators did have was something rarer for a case of this age: a witness who had seen the responsible man directly. Based on Anderson’s account, investigators produced a rough sketch within days. A white man, about six feet tall, behind the wheel of a small light-colored car. For a case of this vintage, that level of detail was unusual.

Most murder investigations of that era had no face to work with at all. On February 19, four days after Mary was found, a vehicle matching that description appeared at a local repair shop. Deputies searched it and questioned the owner when he returned. Nothing connected him to Mary or to the crime scene.

He was cleared. The strongest lead the case produced in its first week led nowhere. Over the following months, the Dunn County Sheriff’s Office worked through more than 400 separate leads, some coming from as far away as Canada. Investigators matched names against Anderson’s description, tracked down similar vehicles, and questioned anyone whose story connected to the afternoon of Mary’s disappearance.

None of it produced a name. What remained after exhausting those 400 leads was the same evidence investigators had started with: a sketch, a description of a car, and a wool cap that no forensic method of the time could use. Years passed. The case did not close, but it went quiet.

The original investigators retired or moved on, taking their direct knowledge with them. New detectives inherited the file, working from papers rather than memory, and hit the same wall the original team had faced. Meanwhile, forensic science continued to advance. DNA testing, unavailable in 1974, became a standard tool in murder investigations during the 1980s and 1990s, helping solve cold cases elsewhere.

For Dunn County, that progress meant nothing without a clear suspect. Mary’s family lived with the same unresolved distance. Thirty-five years passed without an arrest, without a trial, without an answer about why she never reached Chicago. By 2009, investigators reached a critical point.

They contacted Mary’s family with a request no one would have expected thirty-five years earlier: permission to exhume her remains, hoping new testing methods could reveal what the original evidence could not. The family agreed. The exhumation produced results investigators had waited decades for. Detective Scott McRoberts of Dunn County later publicly confirmed that two distinct genetic profiles had been extracted through the new testing.

It was the first genetic material connected to the case since 1974. Investigators compared the new profiles against law enforcement DNA databases. There were no matches. The case needed a suspect who was already under investigation or in the system, and neither existed yet.

They now had biological material directly linked to Mary’s killer. What they lacked was a person to match against it. Two years later, in 2011, that forensic work pointed investigators toward a name with weight far beyond Dunn County: Randall Woodfield. By that time, Woodfield was serving a 90-year sentence, having been convicted as the man nationally known as the I-5 Killer, linked to a series of attacks along the West Coast in the early 1980s.

Before that life defined him, he had been drafted by the Green Bay Packers. Investigators had legitimate reasons to scrutinize him. It is important to be precise about what this was. Woodfield was never charged in Mary’s case.

He was considered a person of interest based on evidence that genuinely pointed in his direction. His documented movements placed him traveling from Portland toward Green Bay at the time of Mary’s death, a route that could have passed through western Wisconsin. His known victim pattern matched what investigators knew about Mary: young women he often encountered while hitchhiking. One detail never quite fit.

Anderson had described the vehicle in 1974 as gold or orange, possibly a Chevy Vega. Years later, some began to wonder whether the color he remembered was closer to light green. Memory, even when vivid, changes over time. That was part of why investigators could not move forward on the description alone.

DNA testing settled the question. Woodfield’s genetic profile did not match the evidence found in the case. He was formally excluded. Because he had never confessed to any of the crimes he was convicted of, Dunn County did not pursue a direct interview with him.

The theory that had seemed stronger than any previous evidence turned out to be the wrong path. The case went cold again for another decade. The real problem became clear. Traditional DNA comparison is powerful when investigators have a suspect or when a matching profile appears in a law enforcement database.

In Mary’s case, neither option had led anywhere. First, there was no way to analyze the evidence. Later, the available testing excluded the only real lead to emerge in decades. The case did not need a stronger test.

It needed an entirely different way of using the available DNA. In 2022, Dunn County partnered with an organization that had opened its doors just months earlier: the Investigative Genetic Genealogy Center at Ramapo College in Mahwah, New Jersey, under the direction of David Gurney. This method works in the opposite direction of traditional database searches. Analysts took the genetic profile extracted from the hair on the cap and uploaded it to public genealogy databases.

The profile did not need to match a suspect directly. It only needed to share segments of DNA with distant relatives who had contributed their genetic data for entirely different reasons. From there, the work became more like systematic research than laboratory science: building a family tree from distant relatives back to a common ancestor, then down through different branches, narrowing thousands of potential relatives into a smaller circle. Two IGG analysts, Karen Binder and Tracy Boyle, were assigned directly to Mary’s case.

The work was slow and careful. But for the first time since 1974, the evidence from that cap had a real path forward, one that did not depend on investigators already knowing who they were looking for. Initial results from the family tree pointed to a man living in Wyoming. Investigators contacted him, and he agreed without hesitation to provide a DNA sample.

The comparison came back negative. He was not the person. In most investigations, this might look like another dead end. But genetic genealogy works differently.

Excluding someone does not erase the progress made. It narrows the search, telling analysts precisely where to focus or where to look again. Then the team hit an obstacle that would have stopped any traditional investigation completely. The man at the center of the family tree they were building had been adopted.

The recorded family lineage, the one on paper, was not his biological lineage at all. Everything traced to that point had been built on the wrong line. That was not a mistake. The paper records simply did not reflect his true biology.

The analysts abandoned that documented history and began rebuilding the search through his actual biological relatives, people who had no idea any of this connected to a murder investigation nearly fifty years old. This was the moment the case truly broke open. Not a clue, not a confession, but a team of genealogists prepared to tear down a family tree they had spent months building and start over. Working from the reconstructed tree, the analysts eventually reached a living relative who could advance the investigation: a daughter with a blood connection to the man at the center of the case.

She agreed to provide a DNA sample. The results were compared directly against the genetic profile from the stocking cap. The match was close enough to leave little room for doubt, pointing investigators to her biological father as the source of that evidence. On November 4, 2024, the IGG Center provided a name to Dunn County investigators: John Keith Miller, 84 years old, living in an assisted care facility in Owatonna, Minnesota.

His name had never appeared once in the case file, not as a suspect, not as a person of interest, not even as a name anyone had written down during fifty years of searching. Before investigators contacted him, public records told their own story. In 1959, Miller was convicted of forgery. A second forgery conviction followed in 1960.

Both sentences sent him to the St. Cloud Reformatory in Minnesota. He was released on parole in January 1963 and married that same year. The marriage did not last.

In 1966, his wife filed for divorce. Court records from that filing cited cruel and inhuman treatment as grounds, and the filing was made while Miller was in custody on another matter: an arrest for robbery in Arizona in September 1965, resulting in about thirteen months in prison. By 1967, Miller had moved to California, where records show a petty theft conviction and a separate robbery arrest that was eventually dropped. In 1969, he was convicted of armed robbery after threatening a convenience store clerk with a loaded gun.

That conviction sent him to San Quentin State Prison. He was released on May 5, 1972. Less than two years later, in February 1974, John Miller picked up Mary Schlais while she was hitchhiking outside Minneapolis. After that day, his record went largely quiet.

A drunk driving conviction in 1994 and a conviction for possessing drug paraphernalia and public intoxication in Iowa in 2014. For nearly ten years, he lived in an apartment directly across from the Austin, Minnesota police department without any documented incident connecting him to them at all. This was not a man who had spent fifty years hiding from the law. He was a man the law had simply never connected to this case.

On November 7, 2024, Sergeant Jason Stocker and Investigator Dan Westland traveled from Dunn County to Owatonna to speak directly with John Miller at the facility where he lived. Miller denied any knowledge of Mary Schlais or the events in Dunn County in 1974. At first, he showed no indication he recognized either name. The investigators presented him with the DNA results and a photo of the wool cap.

According to the criminal complaint, his account changed at that point. He confirmed the cap was his, then described picking up Mary while she was hitchhiking, an encounter that turned violent after she refused his advances, then an attempt to conceal what had happened before a passing vehicle forced him to leave the area. He told investigators he had later disposed of the weapon in a swamp near his residence at the time. It was never found.

Miller was arrested that day, more than fifty years after Mary’s death. In the months following his arrest, John Miller agreed to speak from custody, offering a rare look at his own account of the decades after 1974. The moment that affected him most deeply, he said, was not the arrest itself. It was seeing the wool cap again, presented to him as a photo after all those years.

He recognized it immediately, without any attempt to claim it belonged to someone else. He told investigators he had never imagined that piece of clothing would eventually connect him to Mary’s death. He also described how convinced he had been over the decades that he was safe. Shortly after 1974, he remembered reading a newspaper report describing a vehicle that did not match his own car.

Based on that single detail, he concluded, and continued to believe for the next fifty years, that he had gotten away with it. That belief lasted as long as the evidence remained unexamined. It ended the moment the genealogy team traced the family tree to a daughter of his, and a DNA sample confirmed what the stocking cap had been capable of proving all along. John Miller was formally charged with first-degree intentional homicide.

The case moved through the legal system over the following months, reaching the Dunn County Circuit Court on March 27, 2025. District Attorney Andrea Amadon Nudolph represented the state. Miller was defended by attorneys Shelly Tomtsik and Travis Satorius. Miller entered a no-contest plea, which under Wisconsin law is legally equivalent to a conviction without a direct admission of guilt.

The charge carried a mandatory life sentence. Once the plea was accepted, the court had no discretion to consider any lesser penalty. Judge James Peterson issued the sentence directly: life in prison in the Wisconsin state prison system. The court ordered $2,200 in restitution to Mary’s family.

Miller did not speak in court. Nina Schlais, Mary’s niece, spoke afterward on behalf of the family. She described the day as the end of a fifty-year journey filled with loss, perseverance, and finally justice, while acknowledging that no sentence could erase the pain of losing Mary. She emphasized that Mary was not just a victim but a smart, independent woman, a talented artist, an equestrian, a world traveler, and a scholar.

Miller was then transferred to the custody of the Wisconsin Department of Corrections. He is currently serving his sentence at the Dodge Correctional Institution. Even with the conviction, some parts of the case remain unresolved. The weapon was never found, despite Miller’s account of disposing of it decades earlier.

Randall Woodfield remains formally excluded by DNA and was never charged. His earlier status as a person of interest reflected the evidence investigators had at the time, not a proven conclusion. From the beginning, investigators had genetic material linked to Mary’s killer within days of her death in February 1974. For thirty-five years, no method could turn that evidence into a name.

That was the first wall. When testing finally became available, it pointed investigators toward Randall Woodfield, a conclusion built on a genuine timeline match and genuine pattern match, developed honestly from the evidence available then. But it was also wrong. That was the second wall, and in some ways the harder one, because it looked like an answer before it proved not to be.

What actually solved the case was a different use of that same evidence: genetic genealogy, a technique that did not exist in any usable form until December 2022. The wool cap was never the obstacle in this investigation. It had been carrying the answer from the start. John Miller lived fifty years believing that one wrong detail from a newspaper report had kept him safe.

It had not. Investigators simply had not yet developed the tool needed to find him.