Blog
Cross-Examination of Cooperating Witnesses: What to Expect
The cooperating witness is the prosecution’s most effective instrument and its most obvious liability. In federal criminal cases, the government constructs its narrative around individuals who have already admitted their own guilt and agreed to testify in exchange for sentencing consideration. The exchange is not hidden. It is formalized in a written agreement, filed with the court, and available to the defense. What the jury perceives as candor on direct examination is the product of preparation that began months or years before the witness took the stand.
The Cooperation Agreement
Every cooperating witness operates under a written plea agreement. The terms vary, but the architecture is consistent: the witness pleads guilty to specified charges, agrees to provide truthful testimony and information, and receives in return the government’s promise to file a motion under Section 5K1.1 of the Federal Sentencing Guidelines, recommending a sentence below the otherwise applicable range. The promise is conditional. Truthfulness is required, as the prosecution determines it. Completeness is required, again as the prosecution determines it. The witness occupies a position of dependence that no other participant in the proceeding shares.
That dependence is the first thing a cross-examiner will expose. The cooperating witness has not agreed to tell the truth in the ordinary sense of the phrase. The witness has agreed to tell the truth as the government defines it, with sentencing consequences attached to the government’s satisfaction. A witness whose liberty depends on the prosecutor’s assessment of cooperation is not in the same position as a witness who observed something and reported it.
The plea agreement itself is typically admitted into evidence or read aloud during the government’s direct examination. Prosecutors prefer to address cooperation on their own terms, before the defense can frame it as concealment. The jury hears about the arrangement in the prosecution’s language, which tends toward the clinical: “the witness has agreed to cooperate” and “the government has agreed to recommend a reduced sentence.” The emotional reality of that arrangement, what it means to sit in a cell and calculate the difference between twelve years and three, is left for cross-examination to supply.
What the Government Must Disclose
Under Giglio v. United States, the prosecution bears a constitutional obligation to disclose all promises, inducements, and benefits provided to a witness in exchange for testimony. The obligation extends beyond the formal plea agreement. If the government has made informal assurances, if agents have communicated expectations about the scope or content of the testimony, if there are side agreements or understandings not reflected in the written document, those must be disclosed. A failure to disclose is not a tactical decision; it is a due process violation, and it has been since 1972.
The Supreme Court reaffirmed the force of this principle in February 2025, when it reversed a capital conviction in Glossip v. Oklahoma. The prosecution’s cooperating witness had testified falsely about his psychiatric treatment and medication history, and the prosecutor failed to correct that testimony despite possessing notes confirming the witness had disclosed the truth in a pretrial interview. The conviction had stood for decades. The witness’s testimony was the sole direct evidence connecting the defendant to the murder.
A single failure to correct a single false statement on what seemed a collateral matter was sufficient to require a new trial. The witness was prescribed lithium for bipolar disorder. He told the jury he had received it for a cold.
Glossip is instructive not because it represents an unusual prosecutorial failure but because it illustrates the ordinary vulnerability of cooperating witness testimony. The witness in that case changed material aspects of his account between the first and second trials, denied facts the prosecution knew to be true, and remained the government’s central evidence throughout. These are structural features of the system that produces this kind of evidence.
The Architecture of Cross-Examination
Cross-examination of a cooperating witness proceeds along several lines, though the order and emphasis will vary depending on the specifics of the case, the personality of the witness, and the composition and apparent disposition of the jury (which matters more than most practitioners acknowledge).
The first is motive. The cooperator has a reason to testify that has nothing to do with the truth of the testimony. That reason is the sentencing benefit. A cross-examiner does not stop at establishing that the benefit exists; the jury already knows that from the direct examination. The work is to make the jury perceive the weight of the benefit: the difference, in years, between the sentence the witness faces with cooperation and the sentence the witness faces without it. In narcotics cases, that difference can be the distance between a mandatory minimum of ten years and a sentence of time served. In fraud cases, the gap between a guidelines calculation in the hundreds of months and a recommendation for probation. The cooperating witness knows these numbers. Whether the witness admits to knowing them is one of the small contests of the cross-examination.
The second is preparation. Cooperating witnesses meet with prosecutors and agents before trial, sometimes over the course of months. They review documents. They are told what questions will be asked on direct. They are, if we are being precise, not told what answers to give, but they are informed what subjects will be covered and what documents will be shown to them, which amounts to something adjacent. The defense will ask how many times the witness met with the government, for how long, and what was discussed. The witness will minimize the number and duration of these sessions. In three federal cases we tried in consecutive years, the cooperating witness in each initially described meetings with the government as two or three brief conversations, and in each case the government’s own records reflected seven or more sessions spanning several months.
The third is prior inconsistency. Cooperating witnesses provide multiple accounts over time: to law enforcement during initial interviews, in proffer sessions, before the grand jury, and at trial. These accounts are seldom identical. The inconsistencies range from trivial to material, and the distinction between the two is itself a subject of dispute between the parties. Cross-examination exploits the gaps. A witness who told the FBI in an initial interview that the defendant was not present at a particular meeting, and who now testifies that the defendant was present, must explain the change. The explanation is almost always the same: “I was not being truthful at that time.” The task is to force the jury to consider which version of events deserves less scrutiny.
And the fourth, the one that receives less attention than it should, is the scope of the cooperation itself. What did the witness provide beyond testimony in this case? Did the witness cooperate in other investigations? Against other individuals? Were there cases where the witness’s information was found to be unreliable, or where the government, at any point, considered terminating the cooperation agreement? These questions are not always answerable from the available discovery. The Department of Justice requires prosecutors to review informant files for discoverable information, but the defense does not receive the complete file. What remains undisclosed is, by definition, unknown to us.
Whether that gap in disclosure is meaningful in any given case is a question I cannot answer in the abstract. That is part of the problem.
When the Testimony Shifts
A cooperating witness who has testified in more than one proceeding, or who has given statements at different stages of the same investigation, will have produced a documentary record of the witness’s own account. That record is the cross-examiner’s primary material.
In 2019, before the current wave of federal sentencing reform altered some of the incentive structures around cooperation, the pattern in organized fraud cases was consistent: the cooperating witness’s initial proffer would contain details that served the government’s theory, and subsequent proffers would fill gaps the prosecution had identified. Each successive statement was more complete and more aligned with the charges. The question under cross-examination, stated without ceremony, is whether the story grew because the witness was remembering more or because the witness understood what the government required.
When a cooperating witness is confronted under cross-examination with a prior statement that contradicts the trial testimony, the moment carries weight. The witness pauses. The prosecutor does not object, because there is no valid objection. The jury observes this. What follows depends on preparation and personality. Some witnesses concede the inconsistency and offer an explanation. Others deny that the prior statement means what its plain language conveys. None of these responses eliminate the inconsistency.
The Court in Glossip confirmed what practitioners have long observed: even small inconsistencies in a cooperating witness’s account can be material when that testimony is the primary evidence of the defendant’s involvement. The Napue standard holds that a new trial is warranted if false testimony could in any reasonable likelihood have affected the jury’s judgment. The standard asks whether the lie might have mattered, not whether it did.
Evaluating the Cooperating Witness Before Trial
Most of the work in a cooperating witness case occurs before anyone enters the courtroom.
The defense reviews the plea agreement, the proffer statements, the grand jury testimony, the Jencks Act material, and whatever Giglio disclosures the government has produced. From these documents, the defense constructs a chronology of the witness’s account and identifies every point of divergence. The process is less orderly than that description suggests. Proffer statements are often summaries that agents prepared, not verbatim transcripts. The witness may dispute the accuracy of the summary. The grand jury testimony may be incomplete because the questioning was narrow. The Giglio disclosures may reflect the government’s assessment of materiality rather than a complete accounting of the witness’s background, mental health history, prior dishonesty, and relationships with the individuals under prosecution.
Our practice is to request all records pertaining to the witness’s cooperation in other matters, including cases in which the witness’s information was not used or was determined to be unreliable. The government does not always produce these records without prompting. We submit the request in writing, early in the case, and file a motion if the production is incomplete or delayed. A cooperating witness who has been found unreliable in a prior matter is impeachable on that basis, and the withholding of that information constitutes a Giglio violation regardless of the government’s characterization of the witness’s overall credibility. The standard approach, which treats the government’s initial disclosure as sufficient, overlooks how often material appears in supplemental productions that was absent from the first.
Something like half the time, in our experience, the government’s initial Giglio disclosure in cooperating witness cases contains less than what a complete review of the witness’s file would reveal. I am less certain about that proportion than the sentence implies; it comes from our own caseload and is not a figure anyone has studied with rigor.
What the Jury Instruction Permits
At the close of evidence, the court instructs the jury on how to evaluate cooperating witness testimony. The instruction, in most federal districts, directs jurors to consider the testimony with greater caution, to examine the witness’s motive for testifying, the extent of any benefits received, and whether the testimony is supported by other evidence. The instruction acknowledges that a cooperating witness has an interest in the outcome.
The instruction does not, however, tell the jury what weight to assign. In most federal circuits, a conviction can rest on the testimony of a single cooperating witness, provided the jury believes that testimony beyond a reasonable doubt. Some states require corroboration of accomplice testimony as a matter of law. The federal system does not. The witness may be biased, inconsistent, and self-interested, and the jury is still permitted to convict on that testimony alone.
The Conversation That Begins Before the Courtroom
If you are facing charges in a case where a cooperating witness will testify, the months before trial constitute the most consequential period. The first letter from a federal prosecutor, or the first call from an agent, arrives without reference to who will testify against you. You discover the identity of the cooperating witness later, often in the discovery, sometimes not until the government files its witness list. By that point, the witness has been cooperating for months, occasionally years, refining an account that the defense must now disassemble under the constraints of a trial schedule the court has already set.
A consultation with our firm is where this preparation begins. There is no charge for the initial conversation, and it presumes nothing beyond a recognition that cooperating witness testimony does not answer itself. The question cooperating witness testimony raises must be answered in the courtroom, and the preparation for that answer begins well before the courtroom exists as anything more than a date on a calendar.