It was the spring of 1991, and Grand Junction, Colorado, was in a panic. The first pipe bomb went off on Valentine's Day in the local convention center's parking garage, seriously injuring one person. The second exploded in the wheel well of a family's van and killed a 12-year-old girl in the back seat. A 43-year-old man picked up the third bomb in a restaurant parking lot, and it detonated and killed him instantly.

The bombings left detectives puzzled for months, until James Genrich, a handyman in his late 20s, walked into a local book store and asked about "The Anarchist's Cookbook" — an infamous volume of recipes for bombs, hallucinogens and other illicit items. The store owner called the police, who ultimately searched Genrich's boardinghouse room and seized his tools.

An agent with the federal Bureau of Alcohol, Tobacco and Firearms compared Genrich's tools to markings on fragments of the three exploded bombs as well as a fourth, undetonated device discovered at a motel two years earlier. The agent, John O'Neil, would later testify at Genrich's trial in 1993 that the tools matched "to the exclusion of any other tool in the world."

Jurors convicted Genrich of first-degree murder and other felonies, and the court sentenced him to life without parole. But in the decades since, the widely used forensic toolmark analysis used to convict him has become seen by mainstream science as imprecise and limited at best. Colorado courts recently ruled O'Neil's testimony was inadmissible and ordered a new murder trial for Genrich, who has served about 35 years in prison.

Genrich's case has drawn national attention to the questions surrounding these kinds of forensics. The fates of thousands of defendants have hinged on toolmark and gun analysis, and O'Neil alone had testified as an expert in at least 465 cases, according to court documents.

Chris Fabricant, Genrich's counsel at the Innocence Project, said the reversal of his client's conviction in this case shows the courts finally recognized it was built on a foundation of "junk science."

Meanwhile, even as Mesa County, Colorado, prosecutors conceded O'Neil had overstated the reliability of toolmark forensics, they said they stand by the conclusions he reached. They dropped the murder charges against Genrich, acknowledging a number of practical limitations in conducting a retrial over three decades later. Notably, O'Neil is now elderly and cognitively impaired.

Although only circumstantial evidence against Genrich remains, prosecutors said he is the only suspect in the case — and they still believe him to be guilty.

"While defense experts criticized the field's foundational validity, the people presented testimony from three toolmark experts that the core premise of the science remains valid: Tools leave unique marks, and when trained examiners declare a match, the rate of false identification is very low," prosecutors argued in a brief.

Underlying toolmark and firearms forensics is a simple and seemingly intuitive theory that goes back to the 1800s. The presumption is that no two tools or gun barrels are exactly alike on a microscopic level, so the marks they leave are unique, like fingerprints. A pry bar used in a burglary, for instance, would leave impressions and striations on a window frame that could later be matched to that specific pry bar. Similarly, a gun subtly etches its one-of-a-kind signature into each bullet it fires.

It was not uncommon for forensics experts to make powerful assertions like O'Neil did during Genrich's trial — that no other tool in the world could have made a certain mark. And these bold claims faced relatively little formal scientific scrutiny in the 19th and 20th centuries.

Yet serious legal attacks on the fundamental credibility of toolmark and gun forensics came after the U.S. Supreme Court's 1993 landmark decision in Daubert v. Merrell Dow Pharmaceuticals Inc. , which set stricter standards for admitting scientific expert testimony at trial and established judges as gatekeepers. Justices handed down the Daubert opinion 47 days after Genrich's sentencing.

Daubert opened new avenues for attorneys to exclude certain types of forensic analysis, for which there was relatively sparse empirical research in the scientific mainstream.

In 2009, the National Academy of Sciences published a 350-page report assessing the state of forensic sciences. It found that leaps in DNA technology have helped law enforcement identify culprits and, in some cases, revealed that innocent people have been wrongfully convicted based on bad forensics.

"This fact has demonstrated the potential danger of giving undue weight to evidence and testimony derived from imperfect testing and analysis," the report said. "Moreover, imprecise or exaggerated expert testimony has sometimes contributed to the admission of erroneous or misleading evidence."

The report zeroed in on "individualization," where a forensic analysis purports to match evidence to a specific source — a toolmark to a tool, a bullet to a gun, or a bite mark to a set of teeth. Unlike DNA analysis, none of these methods were highly reliable or consistent, and standards among forensics experts often differed, the report said.

"The simple reality is that the interpretation of forensic evidence is not always based on scientific studies to determine its validity," the report said. "This is a serious problem. Although research has been done in some disciplines, there is a notable dearth of peer-reviewed, published studies establishing the scientific bases and validity of many forensic methods."

Federal courts, however, had not consistently ensured that forensic testimony was scientifically valid and reliable as the Daubert decision required — and rarely excluded or restricted the testimony of prosecutors' experts, according to the report.

"The adversarial process relating to the admission and exclusion of scientific evidence is not suited to the task of finding 'scientific truth,'" the National Academy of Sciences report said.

In 2016, the U.S. President's Council of Advisors on Science and Technology issued a report calling for clear standards for the validity and reliability of forensic methods, and an evaluation of forensic methods to ensure each is scientifically valid and reliable.

The report raised serious concerns about non-DNA-related forensic analyses, such as latent fingerprints, bite marks, footwear and toolmarks. For instance, the group concluded that matching footwear impressions left at crime scenes to specific shoes was "unsupported by any meaningful evidence or estimates of their accuracy, and thus are not scientifically valid."

Few valid black-box studies had been conducted to measure the rates of false positives in the fields, the report found. The studies that did exist suggested significant rates of error. For example, a black-box study of firearm analysis — a close cousin of toolmark analysis — found a false-positive rate of 1 in 66. The report recommended presenting such forensic analysis with a clear disclosure of the technique's error rates.

The year the President's Council of Advisors on Science and Technology's report was released, Genrich challenged his conviction based on new evidence, saying O'Neil's evidence was no longer supported by mainstream science. His motion included an affidavit from forensic scientist Jay Siegel, who was on the National Academy of Sciences committee in 2009 that found toolmark examiners had overstated their ability to match marks to individual tools.

The district court initially denied his motion without a hearing, saying Genrich needed to prove the new evidence "neutralized" O'Neil's testimony — yet the new evidence only impeached it. An appellate panel reversed the decision and ordered the lower court to grant Genrich a hearing.

That hearing turned out to be a battle of experts — three for the defense, who attacked the validity of toolmark analysis generally and in Genrich's case; and three for prosecutors, who defended it.

University of California College of the Law, San Francisco Dean David Faigman was among the experts who testified that toolmark analysis is not a valid science. He pointed to studies showing firearm toolmark examiners' error rates ranged from 20% to 54% if inconclusive results were counted as incorrect.

Faigman and other experts testifying for Genrich told the court that analyses of marks made by handheld tools were even less reliable than analyses of bullets and guns — bullets only travel down gun barrels one way, while tools can be gripped and pulled and used in many different directions, making an accurate analysis much more difficult.

Prosecutors' firearm and toolmark experts, meanwhile, testified they had never made an incorrect match, according to court documents.

In a recent interview, Faigman told Law360 he has given testimony over the past seven years in about 40 cases involving the admissibility of forensics. Most of the cases are new, he said.

"In a relatively short period of time over the last six, seven years, there has been a sea change in how courts see firearms and toolmarks, which obviously, in my opinion, is a very good outcome," Faigman said. "There are just a lot of areas of forensic science that we have now moved away from because we know that they were not valid, and they should never have come into court in the first place, and that's essentially true for firearms and toolmarks."

Michael Haag, president of the Association of Firearm and Toolmark Examiners, told Law360 his field is growing and evolving, yet he defended its validity. He said the Genrich case does not call into question the scientific underpinnings of toolmark forensics, yet its outcome brings into focus a change in how formal findings are presented — forensic examiners no longer use the phrase "to the exclusion of all others."

"Like all scientific endeavors, firearm and toolmark analysis is a dynamic field," Haag said. "The shift away from 'exclusion of all others' represents the natural and expected progression of a discipline refining its communication of observational results."

He added that the court's decision to order a new trial for Genrich "was not a Daubert challenge regarding the validity of the science itself, but rather a focus on the outdated terminology used in the original testimony."

O'Neil, the ATF agent who analyzed Genrich's tools, had testified in the 1993 trial that three of the handyman's tools had been used to create the four bombs in Grand Junction: Genrich's needle-nosed pliers cut two wires on the undetonated 1989 bomb, his slip-joint pliers left impressions on fragments recovered from the two fatal explosions, and his wire cutter snipped a wire recovered from another.

However, O'Neil did not take notes or measurements, nor did he write a report about the matches that analysts could verify today, according to court documents.

In 2023, the district court found that O'Neil's assertion that Genrich's tools matched to the exclusion of all other tools was no longer admissible based on current scientific standards, granting Genrich a new trial. Prosecutors appealed, but an appellate court panel upheld the decision in May 2025, and the Colorado Supreme Court declined to take the case.

In April of this year, prosecutors agreed to drop the two murder charges against Genrich before his retrial, saying O'Neil would not be able to testify.

"We explored new forensic and investigative leads, but current science has not been able to replicate or replace the specific identification testimony excluded by the court," Mesa County prosecutors said in their motion to dismiss the murder charges.

Prosecutors added that "O'Neil was a dynamic witness who intertwined his testimony with various physical demonstrations. It would be extremely difficult, if not impossible, to recreate those visual demonstrations from a cold paper record."

Last year, prosecutors sent bomb fragments to a respected independent laboratory to use modern toolmark analysis methods, but the lab returned inconclusive findings, according to the dismissal motion.

They also submitted evidence from the case for modern fingerprint and DNA testing, yet no DNA was found. Analysts did, however, discover an unidentified fingerprint that did not belong to Genrich.

At least 28 witnesses who testified in Genrich's trial have died in the past 35 years, prosecutors also noted.

However, they said they still have circumstantial evidence pointing to Genrich as the primary suspect, and no other viable suspects have been identified.

For example, police had found angry notes in his room describing plans or fantasies of killing people, as well as references to the bombings themselves, which were written weeks afterward. He directed particular anger at "stuck-up" women who spurned him romantically.

A local business employee testified that Genrich threatened to "kick someone's ass or kill somebody," believing that another employee had gotten him fired from a job at the Two Rivers Plaza in Grand Junction, where one of the bombs later detonated.

He also had skills that would enable him to build such devices, prosecutors said, and he lived within walking distance of a hardware store that was the only business in the region that carried one of the main bomb components. He was also seen in the Two Rivers Plaza hours before a bomb exploded there.

But Genrich was living and working in Phoenix when the first, unexploded bomb was planted in 1989. And Genrich's parents testified he had been with them at their house when each of the other three bombs had been placed in 1991.

On balance, the courts found the circumstantial evidence was not enough to link Genrich to the bombs without O'Neil's now-inadmissible testimony.

While prosecutors begrudgingly dropped the murder charges, Genrich's convictions and 72-year sentence for explosive device and assault charges still stand. Fabricant, Genrich's counsel with the Innocence Project, said the district court had denied a motion to dismiss those counts because prior counsel failed to argue for an equitable tolling of the statute of limitations to challenge the lesser counts.

"It's incredibly frustrating and profoundly unjust that Mr. Genrich remains incarcerated on what the courts have all agreed is debunked evidence," he said.

Because courts dismissed his murder convictions, Genrich is now eligible for parole. A hearing is scheduled for this week, Fabricant said.

Prosecutors plan to oppose Genrich's release.

Mesa County Assistant District Attorney Jeremy Savage told Law360 in a recent interview that this should not be seen as an exoneration, as prosecutors still believe Genrich was the bomber and should be serving life without parole.

"I know the defense, you know, the Innocence Project, is kind of taking the position that they believe that he is in [prison] on a technicality," Savage said. "Well, we believe it's the opposite. We believe that the reason that he is potentially parole-eligible in this case now is on a technicality."

The state is represented by Daniel P. Rubinstein, Jeremy Savage and Michael Fisher of the Mesa County Colorado District Attorney's Office.

Genrich is represented by M. Chris Fabricant and Marika Meis of the Innocence Project, Brian Liegel, Greg Silbert, Irwin H. Warren, Corey Brady, Marina Masterson, Anastasia Zaluckyj and Jill Jacobson of Weil Gotshal & Manges LLP, Kathleen Lord of the Korey Wise Innocence Project, Scott Troxell, and Rebekka Higgs.

The case is People of the State of Colorado v. Genrich, case number 92CR95, in the 21st Judicial District of Colorado.

--Additional reporting by Elizabeth Daley. Editing by Philip Shea and Lakshna Mehta.

Have a story idea for Access to Justice? Reach us at accesstojustice@law360.com.