Til Death Do Us Part
Two Bodies Made a Survivor’s Account Difficult to Ignore
Thomas Randolph called the police with an explanation for the two people lying dead in his home. An intruder had killed his wife. Randolph had then killed the intruder. If that account was correct, he was a husband who had survived a violent attack and stopped the person responsible.
The scene in Las Vegas on 8 May 2008 required investigators to test that explanation against the bodies, the house and Randolph’s own movements. They also needed to identify the supposed stranger. That identification opened a connection which made the night harder to understand as a simple confrontation between a homeowner and a burglar.
Netflix’s documentary The Widower: ’Til Death Do Us Part returns to the case through the work of journalist Dan Slepian and detective Dean O’Kelley. Its central question is how an apparently self-contained account of a home invasion became an investigation stretching through relationships, insurance, earlier allegations and years of legal argument.
What Randolph Said Happened Inside the House
Randolph said he and his wife, Sharon Causse, had returned home after an evening out. Sharon entered the house while he remained in the garage with their vehicle. When he subsequently went inside, he found her lying face down in the hallway.
According to his account, unexpected movement alerted him to another person. He obtained a handgun, confronted a masked intruder and struggled with him before firing. The man collapsed in the garage, where Randolph fired two further shots into his head.
The injuries were specific. Sharon had died from a single gunshot wound to the head. The other person had sustained five gunshot wounds, including two to the head. Those findings established that both deaths involved gunfire, but the number of wounds alone could not explain the intentions, sequence or legal justification behind every shot.
Randolph’s version supplied those missing connections. He placed himself outside the house when Sharon was attacked, described discovering the aftermath, and presented the second shooting as a response to danger. Each part created something investigators could examine: where he had been, what he could have seen, where he obtained his weapon and how the encounter moved through the property.
A survivor’s account is indispensable in such circumstances, but it is not self-verifying. The person giving it may be the only living witness to important moments. That makes the comparison with independent evidence more necessary, not less.
The man in the garage was Michael Miller. He was not simply an unidentified intruder whose connection to the home began that evening. Randolph knew him, and their association became a central part of the inquiry.
The Connection Behind the Mask
Miller had done odd jobs for Randolph, who had also lent him money. The two men communicated frequently in the months before the deaths. The appellate record describes hundreds of calls and an association that included private conversations.
They had also been together only hours before the shootings. That did not, by itself, establish a criminal agreement. People can know someone who later commits a crime against them. But it made an explanation based on an entirely unexpected stranger inadequate.
Investigators now had to examine the relationship as well as the physical encounter. What did Miller know about the household? What had he been told about that evening? Did his presence fit an independent burglary, or did it connect to a plan involving someone inside the home?
Phone records could establish contact, not the words exchanged. That limitation is important. A large number of calls may contradict an attempt to minimise a relationship, but a call log does not contain an agreement to kill. Its evidential force depends on the other material with which it is considered.
The same applies to financial help. A loan or job can explain why two people speak regularly. Prosecutors would need more than familiarity to prove conspiracy. The investigation therefore had to move from the existence of a connection to the significance of that connection on the night Sharon died.
Miller’s relatives supplied another element. According to evidence later summarised by Nevada’s Supreme Court, he had said shortly before the killings that he expected money from Randolph and would be leaving the state. Those statements mattered because they concerned anticipated events, rather than an explanation invented after his death.
They remained witness evidence about what Miller had said. They needed to be assessed for reliability and context, and they did not reveal every detail of what he believed would happen. Their importance lay in how they fitted, or failed to fit, the rest of the emerging picture.
What the House Could Tell Investigators
Police found no signs of forced entry. Near Miller’s body was a bag containing costume jewellery and keys to the home. A gold ring was found in his pocket. These details complicated any easy distinction between a genuine theft and a scene arranged to resemble one.
The absence of forced entry does not prove that a person was invited. A door may be unlocked, or access may have been obtained by other means. But keys connected to the property raise a further question about opportunity and access, especially when the person carrying them has a relationship with the homeowner.
Jewellery can also support more than one interpretation. It may indicate that someone was stealing. It may also form part of an account in which theft provided cover for another purpose. The presence of stolen-looking property cannot, on its own, settle whether a burglary was independent, planned jointly or staged.
This is why the case cannot be reduced to a single dramatic clue. The physical items needed to be considered alongside the relationship, the timing, the statements to relatives and Randolph’s account of his own actions. A coherent explanation had to accommodate the whole combination.
Investigators also examined discrepancies between his version and the physical evidence. The published judicial account records those inconsistencies as part of the basis for suspicion, but it does not justify inventing a precise forensic reconstruction beyond what the record establishes. A confident diagram or imagined sequence would not become evidence merely because it looked persuasive.
The investigative task was narrower and more demanding: identify which statements could be checked, determine which were supported, and assess whether the unsupported parts were mistakes, limitations of recollection or indications of a false account.
Why the Insurance Policies Mattered
Randolph had taken out multiple life-insurance policies on Sharon. Prosecutors would argue that the prospect of collecting the proceeds supplied a motive for arranging her death. The alleged financial benefit made the relationship with Miller and the circumstances of the shooting more consequential.
Insurance is nevertheless common and lawful. Its existence cannot establish that a policyholder or beneficiary intends a death. To treat every insured spouse as suspicious would turn an ordinary financial arrangement into an accusation without evidence.
The question was how the policies interacted with the rest of the case. Were they part of a pattern of preparation? Did the expected benefit help explain an otherwise difficult sequence? Could the prosecution demonstrate an agreement and actions directed towards that benefit, rather than simply point out that money was available?
Motive is useful because it can make conduct intelligible, but it is not a substitute for proving conduct. A person may have a possible financial motive and still be innocent. Equally, an established crime does not require investigators to recover every private thought that preceded it.
In Randolph’s case, the insurance theory gave prosecutors a proposed reason for the alleged arrangement. The evidence about Miller was intended to show how that arrangement could have been carried out. The two parts had to support one another without being mistaken for the same thing.
That distinction also explains why a true-crime programme can feel more conclusive than the underlying evidence warrants at a particular point. A financial motive followed by an ominous relationship creates a compelling narrative. A court still has to decide whether the admissible evidence proves the charged offences beyond reasonable doubt.
An Earlier Marriage Enters the Investigation
As detectives examined Randolph’s history, they encountered the death of his second wife, Becky Gault, in Utah in 1986. She had died from a gunshot wound to the head. Her death was initially treated as suicide, but authorities later brought a murder charge against Randolph.
A former friend, Eric Tarantino, alleged that Randolph had tried to recruit him to kill Becky, including through a proposed staged burglary. Tarantino said life-insurance proceeds had been discussed. Those allegations appeared to offer a striking parallel with the theory developing around Sharon’s death.
There was, however, a decisive limit: a Utah jury acquitted Randolph of Becky’s murder in 1989. That acquittal cannot be replaced by the later documentary’s atmosphere or by the resemblance between accusations. He was not convicted of killing her.
The Utah proceedings also involved a separate witness-tampering matter. The later Nevada judgment records that Randolph solicited an undercover officer to kill Tarantino and pleaded guilty to felony witness tampering. The records relating to the Utah prosecution and conviction were subsequently expunged.
These were different legal events. An allegation about Becky’s death, an acquittal on the murder charge and a guilty plea in the separate witness matter should not be compressed into a single claim that an earlier killing had been proved.
The wider fact that Randolph married six times, and that four former wives had died, makes a memorable headline. It does not establish four murders. The individual circumstances and legal outcomes must remain visible, otherwise the arithmetic supplies a conclusion that the evidence has not established.
For investigators, the earlier events were possible leads. For prosecutors, their use raised another question: how much of that history could fairly be put before a jury deciding what happened in Nevada in 2008?
When a Pattern Becomes a Legal Problem
The prosecution sought to introduce evidence from Utah for purposes including motive, intent, preparation and plan. The defence objected. The dispute concerned more than whether the earlier material was interesting or disturbing; it concerned what it could legitimately prove about the charged offences.
Nevada’s evidence rules distinguish between using an earlier act for a relevant purpose and using it to suggest that a defendant is the kind of person who would commit the present crime. The second route risks replacing proof of the specific allegation with condemnation of character.
The distinction can be difficult in practice. A previous alleged plan may resemble a later alleged plan. That resemblance can potentially illuminate intent or method. Yet it can also tempt a jury to punish a defendant for past allegations, especially when the earlier account is vivid and emotionally powerful.
The court must therefore consider both relevance and unfair prejudice. Relevant evidence is not automatically admissible in every form or quantity. Its usefulness can be outweighed by the risk that it will distract the jury, confuse the issues or produce a decision for an improper reason.
The Utah evidence also had a complicated procedural history because the earlier murder charge ended in acquittal. That did not mean every related fact became impossible to discuss in any later proceeding. It did mean that the court needed particular care in assessing what was being proved, through which witnesses and for which purpose.
This became one of the most important parts of Randolph’s case. The public could see a sequence of marriages and deaths. The legal system had to decide which pieces of that sequence belonged in a trial about two particular victims.
The Difference Between a Documentary and a Trial
A documentary can move between decades, interviews, archive footage and personal impressions. It can show a subject speaking at length and invite viewers to judge how convincing he appears. A trial has a different task and different constraints.
The jury must decide the charges using the evidence admitted under the governing rules. It cannot lawfully substitute a general dislike of the defendant for proof. Nor does a person’s apparent charm, arrogance or emotional flatness independently establish what happened inside a house.
That is especially important with Randolph because personality forms part of the documentary’s appeal. Viewers may find his manner revealing. Such reactions can explain how the film affects an audience, but they should not be treated as a psychological diagnosis or a forensic test for lying.
The more reliable questions concern checkable propositions. Does a statement match a record? Does it account for an item found at the scene? Does an explanation survive comparison with another witness’s evidence? These questions can be pursued without assuming that everyone displays fear, grief or innocence in one recognisable way.
The documentary’s long timespan also creates a risk of hindsight. Once viewers know that police developed a theory, earlier behaviour can seem to point inevitably towards it. Investigators did not begin with a finished narrative. They had to decide which leads justified further work and which similarities might prove misleading.
For Sharon and Miller, the investigation concerned their deaths as individual events. Keeping them at the centre prevents the case from becoming merely an exercise in watching an unusual defendant perform for a camera.
Why the Case Took More Than One Legal Answer
The eventual proceedings required courts to address both factual responsibility and the fairness of the process used to decide it. Those are connected questions, but they are not identical. A finding about an evidential error does not necessarily determine who committed a crime.
Likewise, the passage of time does not explain itself. A long case can involve motions, changes of representation, appeals and a retrial. The elapsed years are important, particularly to affected families, but legal responsibility for delay must be examined through the actual procedural record.
The later appeal considered arguments concerning the speed of the proceedings, statements made during a walkthrough of the house, the handling of a computer used to review case material, jury directions and the conduct of closing arguments. Each raised a specific question rather than a general request to decide whether the case felt fair.
The walkthrough issue illustrates the method. Randolph argued that his participation was involuntary and affected by medication. The court considered testimony from the detectives and the recording, including whether he appeared to understand and answer the questions. It assessed the circumstances together rather than treating medication use alone as conclusive.
The computer issue required evidence that privileged material had actually been accessed through improper interference. An allegation of possible intrusion was not enough. The distinction between a concerning possibility and a demonstrated violation shaped the court’s analysis.
These procedural arguments are often shortened in retellings because they lack the immediate drama of the crime scene. Yet they explain why a case can change course after a verdict and why the final legal position must be checked rather than assumed from an older programme or headline.
How to Read a Case Built from Connected Facts
A circumstantial case asks the jury to reason from facts rather than rely on a witness who watched the whole crime happen. That does not automatically make it weak. It does, however, make the connections between those facts especially important. A persuasive account must explain why the evidence supports the alleged sequence and why competing explanations are less convincing.
In Randolph’s case, the relationship with Miller changes the meaning of the apparent intrusion. Contact between two men is not proof of murder. Nor is an insurance policy. Nor is an absence of forced entry conclusive on its own. The prosecution’s argument depends on how these circumstances interact with the deaths and with Randolph’s description of what happened.
Readers can test that argument without pretending to retry the case. Begin by separating the observation from the inference. The observation might be that the men communicated repeatedly. The inference concerns what they were arranging. Keeping those steps distinct prevents a suspicion from becoming an apparently established fact simply through repetition.
Then ask what would be expected under each account. If Miller was acting entirely independently, how does that fit the existing relationship? If a burglary was the purpose, how does the scene fit that purpose? If Randolph’s description was accurate, which details support it and which require additional explanation? These are questions about the coherence of the evidence, not instructions to treat every unusual detail as incriminating.
The same discipline applies to the earlier marriage. Similarities can feel powerful because a repeated story seems to explain a person. Yet a resemblance between two events does not remove the need to prove what happened in either one. The danger is that uncertainty in the earlier event becomes hidden once it is used to explain the later event.
There is also a difference between recognising a possible alternative and demonstrating reasonable doubt. An alternative account needs to be assessed against the evidence rather than protected from it. Equally, the prosecution does not get to fill an evidential gap by pointing to how unpleasant or suspicious a defendant appears in an interview.
For a documentary viewer, this is a useful way to resist the force of editing. A pause, facial expression or unsettling piece of music can guide an emotional response without adding a fact about the killings. The most revealing question is not whether Randolph seems like the sort of person who might commit the crime. It is what the properly admitted evidence establishes about this crime.
That distinction becomes decisive when following the case through its different trials and appeals. The legal history is not an inconvenient interruption to the story. It explains which evidence could be used, how the verdict was reached and what survived judicial review.
What the Courts Ultimately Established
Randolph was convicted in 2017 of conspiracy to commit murder and two counts of first-degree murder with a deadly weapon. He received a death sentence. In December 2020, Nevada’s Supreme Court reversed the judgment and ordered a new trial because the presentation of the Utah evidence created unfair prejudice that the prosecution had not shown to be harmless.
That ruling was not an acquittal of the Nevada charges. It required the state to prove its case again through a legally acceptable trial. The court criticised the breadth and effect of the earlier material, including evidence that went beyond the limited purposes advanced for admitting it.
At the retrial in 2023, a jury again convicted Randolph. The resulting sentence was an aggregate term of sixty years to life. The prosecution’s established case was that he conspired with Miller to kill Sharon during a staged home invasion and then killed Miller.
Randolph continued to deny responsibility. His later appeal did not succeed: an order dated 31 August 2026 affirmed the judgment. Some subsequent coverage described the appellate development in September, but the available order itself bears the August date.
The court rejected the challenges before it, including the arguments concerning delay, the walkthrough and jury instructions. It also found no cumulative error requiring reversal. As of the record checked for this article, the convictions and the sixty-years-to-life sentence therefore remained in place.
The strongest answer to the documentary’s mystery is consequently more precise than its nickname. Randolph was convicted over Sharon Causse’s and Michael Miller’s deaths, after an initial conviction was overturned and a second jury heard the case. Becky Gault’s murder charge ended in acquittal, and the deaths of other wives must not be silently converted into additional convictions.
The case’s enduring lesson lies in the distance between a persuasive story and a proven one. The home-invasion account had to be tested against relationships, objects, timing and financial evidence. The prosecution’s account then had to survive rules designed to prevent suspicion about a person’s past from doing the work of proof. Both forms of scrutiny were necessary to reach the legal outcome that now stands.
Sources
Nevada Supreme Court — Randolph v. State, 10 December 2020 — Judicial opinion reproduced by CaseMine; the scene, the Utah evidence and the reasons for reversal.
Nevada Supreme Court — Order of Affirmance, Case 88625, 31 August 2026 — Judicial order reproduced by CaseMine; retrial, sentence and later appellate findings.
Netflix Media Center — The Widower: ’Til Death Do Us Part — Documentary identity, production and premise.
The National — Who Is Thomas Randolph? — Reporting on the marriages, documentary and procedural chronology.
Next Reads
Lizzie Borden Explained: The Evidence Against Her Ranked — Another case requiring a clear distinction between suspicion and a jury’s verdict.
The Lizzie Borden Story — What Really Happened? — How screen storytelling can alter the meaning of historical evidence.
