Blog
Cooperation Agreements & Plea Deals
Cooperation Agreements and the Illusion of Reciprocity
Cooperation does not produce what most defendants believe it will produce. The word itself suggests mutuality, an exchange of roughly equivalent value, when in reality the federal cooperation agreement is among the most lopsided instruments in criminal practice. One party surrenders information, testimony, safety, and the ability to mount certain defenses at trial. The other party retains the sole discretion to determine whether any of that sacrifice merits recognition.
The mechanism is codified in Section 5K1.1 of the United States Sentencing Guidelines and in 18 U.S.C. § 3553(e), and it operates with a simplicity that obscures its consequences: only the government may move for a downward departure based on substantial assistance. The Supreme Court confirmed this structure in Wade v. United States, and in the decades since, the architecture has not changed.
What has changed is the frequency with which defendants are encouraged to cooperate without a clear understanding of what cooperation will cost or whether the government regards their information as sufficient to warrant the motion that makes the cooperation meaningful. Most of the cases we see involve people who cooperated, who told the truth, and who received nothing for it except a proffer statement the government can now use to calculate a higher offense level at sentencing.
The Proffer Session
Before a cooperation agreement is signed, there is a proffer. The agreement is sometimes called a “queen for a day” letter, a name that communicates more about its nature than the document’s drafters likely intended. The protections it offers are, if we are being precise, not protections at all but limitations on the manner in which the government may use what the defendant provides.
The proffer agreement prevents the government from introducing the defendant’s statements in its case in chief at trial. That exclusion is narrower than most defendants perceive. The government may still use those statements for impeachment if the defendant testifies inconsistently at a later proceeding. It may use the information to develop derivative leads, pursue additional witnesses, and construct a broader theory of the case. Under the federal sentencing guidelines, everything disclosed during the proffer can be treated as relevant conduct when calculating the offense level.
In 2019, before certain Justice Manual provisions were amended, we observed a pattern in the Eastern District that persists: defendants entering proffer sessions without a completed review of their own exposure, providing information that expanded the government’s understanding of their conduct, and receiving in return nothing that resembled a cooperation agreement. The proffer session ended. The prosecution continued.
The proffer letter promises not to use your words against you. It does not promise that your words will not be used.
Whether the distinction matters more in the Second Circuit than in the Eleventh, where the derivative use question has received less attention, is a question that the case law has not resolved to anyone’s satisfaction.
The preparation required before a proffer session is considerable, and it is the area where the quality of representation determines whether cooperation produces a benefit or a liability. The defendant must reconstruct, often across years of conduct, a narrative that is complete enough to satisfy the government’s requirement of truthfulness and precise enough not to expand the defendant’s own exposure inadvertently. A session where the defendant volunteers information about uncharged conduct, in the hope that candor will be rewarded, is a session where the government may discover offenses it had no prior knowledge of. The obligation to be candid is fixed in the agreement. The government’s obligation to reward that candor is not.
Prosecutorial Discretion Under Section 5K1.1
The Supreme Court in Wade established that the government’s authority to file a substantial assistance motion is a power, not a duty. The language is important because it forecloses the argument that cooperation, however complete and truthful, entitles the defendant to a departure.
The only recognized basis for judicial review of the government’s refusal to file a 5K1.1 motion is unconstitutional motive: the defendant must demonstrate that the refusal was based on race, religion, or the exercise of a constitutional right. The Second Circuit, in United States v. Rexach, outlined a somewhat broader framework, permitting review where the government’s refusal constitutes bad faith in the fulfillment of its contractual obligations under a cooperation agreement. The distinction between the Wade standard and the Rexach standard is one we find consequential in practice, though neither is easy to satisfy.
What this means for the defendant considering cooperation is that the prosecutor who charges you is the same person who decides whether your assistance deserves recognition. You provide information. The prosecutor evaluates it. If the prosecutor determines the information was not substantial, or not timely, or not useful enough to the investigation of another person, the 5K1.1 motion is not filed. The court cannot intervene. The defendant serves the guideline sentence, or the mandatory minimum, with no credit for the cooperation that may have cost relationships, safety, and the ability to present a viable defense at trial.
I am less certain than the preceding paragraphs might suggest about whether this dynamic produces worse outcomes than the alternative. A system in which defendants could compel the court to reward cooperation regardless of the prosecutor’s assessment would create its own distortions. The problem is not that prosecutors have discretion; the problem is that the defendant must exercise irreversible trust in a party whose interests remain, in every other respect, adversarial.
The interplay between Section 5K1.1 and 18 U.S.C. § 3553(e) compounds the difficulty. A 5K1.1 motion permits the court to depart below the guideline range. But if the defendant faces a statutory mandatory minimum, the 5K1.1 departure alone cannot breach that floor. A separate § 3553(e) motion is required, and that motion lies within the prosecutor’s sole discretion as well. We have seen this produce results that are difficult to reconcile with the stated purpose of cooperation: three cases in the past eighteen months, all involving mandatory minimums of ten years or more, where the government filed the 5K1.1 motion but declined to file the § 3553(e) motion. The defendant who cooperated received a sentence at the mandatory minimum floor.
The guidelines direct the court to consider five factors when determining the extent of a departure: the significance of the assistance, its truthfulness and reliability, the nature and extent of the cooperation, any risk to the defendant or the defendant’s family, and the timeliness of the assistance. The court considers these factors after the government has already decided to file the motion. The evaluation concerns degree, not eligibility. Eligibility remains the prosecutor’s determination.
What Cooperation Demands in Practice
The image most defendants carry into the first meeting with counsel is of a single conversation: tell the government what you know, receive credit for it at sentencing. The reality is that cooperation is an obligation that may extend for years, encompassing debriefing sessions, grand jury testimony, trial testimony, and, in some cases, activities that place the defendant in proximity to the ongoing criminal conduct of others.
A debriefing is an interrogation conducted under the terms of a proffer agreement, led by agents with access to the government’s full investigative file. The defendant’s attorney is present but cannot direct the session. The agents ask about every person the defendant has interacted with, every transaction, every communication. The truthfulness obligation extends not only to the questions asked but to the information the defendant knows to be relevant and does not volunteer.
What the defendant experiences during this period is the psychological weight of performing obligations for the entity prosecuting you, under conditions where any omission, whether the product of dishonesty or the natural degradation of memory across time, can void the agreement entirely. The government’s standard for truthfulness is absolute. Memory is not. That gap is where cooperation agreements most often fail, not because the defendant lied, but because the defendant remembered imperfectly, and the government interpreted the imperfection as deception.
We approach proffer preparation differently than the process most defendants have described to us from prior representations. Before any session, our practice requires an internal debriefing of the client that mirrors the government’s anticipated scope. The client reconstructs the relevant conduct in a structured chronology. We identify the areas where the client’s memory is uncertain and distinguish those from the areas where the client’s knowledge is incomplete. The distinction matters because the government treats “I do not recall” and “I do not know” as different statements, and conflating them during a proffer session can produce the appearance of evasion where none exists.
When Not to Cooperate
Not every defendant has information the government values. The structure of federal conspiracy cases ensures that the individuals with the most exposure, the lowest level participants and the employees who followed instructions without understanding the scope of the operation, are often the individuals with the least to offer. A defendant who cannot identify co-conspirators or describe the organizational structure possesses nothing that would satisfy the substantial assistance standard.
For that defendant, cooperation is not a strategy but a proffer session that produces nothing except a record of the defendant’s own admissions, admissions that may increase the sentencing guideline calculation without generating any corresponding motion for departure.
The question one must answer before the first proffer session, the question that determines whether cooperation serves the defendant’s interest or the government’s, is whether the defendant possesses information the government does not already have. If the answer is uncertain, the proffer may confirm what the government already knows without providing anything it would characterize as substantial. The defendant cooperated, the government agrees the defendant cooperated, and no 5K1.1 motion is filed because the cooperation, while truthful, did not advance the investigation of another person.
Cooperation agreements occupy a position in federal practice that few other instruments share: they require the defendant to trust the adversary, to relinquish defenses before knowing whether the exchange will produce any benefit, and to perform obligations that may extend years beyond the signing of the agreement, all under the condition that the adversary retains unilateral authority to determine whether the performance was sufficient.
The February 2025 policy memorandum from the Attorney General’s office, which instructed federal prosecutors that charging and plea decisions must not be influenced by political considerations, reaffirmed the existing framework without altering the distribution of power within it. The memo’s caution against using criminal charges to exert plea leverage is reasonable in principle, though in the cooperation context, the asymmetry that existed before the memo persists.
A consultation is where the evaluation of that asymmetry begins. Before any proffer is scheduled, before any agreement is signed, the question that requires an answer is whether the defendant’s information possesses sufficient value to justify the irreversible disclosures cooperation demands. That question cannot be answered by the defendant alone, and it cannot be answered by a firm that does not understand, from repeated experience, how the government in a particular district evaluates what it receives.