United States v. Cardenas: Corroborative Third-Party “Other Acts” Evidence Is Not Rule 404(b) Propensity Proof When Offered to Show Defendant’s State of Mind

Court: U.S. Court of Appeals for the Second Circuit Date: February 18, 2026 Docket: 24-2734

1) Introduction

United States v. Cardenas concerns a Colombian patrol-level police officer, Jey James Roldan Cardenas (“Roldan”), convicted after a jury trial in the Southern District of New York (Kaplan, J.) of conspiracy to import cocaine into the United States, in violation of 21 U.S.C. §§ 963 and 960(b)(1)(B). The government’s evidence showed Roldan participated in meetings and communications with Isidro Vargas, a DEA confidential informant posing as a trafficker, about moving large quantities of cocaine through the Cartagena airport.

Roldan’s defense did not deny the communications. It denied criminal intent. Roldan testified he believed he was helping a fellow Colombian National Police (“CNP”) officer, Jose Alfredo Aguas Oviedo (“Aguas”), set up a law-enforcement seizure of cocaine—an operation that would generate reward money—rather than joining a trafficking conspiracy.

The appeal turned on a discrete evidentiary ruling: the district court excluded undisputed evidence (via an unavailable CNP Major’s statement) that Aguas had twice previously provided information leading to successful cocaine seizures. The Second Circuit held that exclusion was error, not barred by Rule 404(b), and not harmless because it went to the heart of intent and credibility.

2) Summary of the Opinion

The Second Circuit (Robinson, J., joined by Carney and Park, JJ.) vacated Roldan’s conviction and remanded. The court held:

• The excluded evidence was relevant under Rule 401 because it corroborated Roldan’s account of key conversations with Aguas that informed Roldan’s stated belief and thus his intent.
• The evidence was not barred by Rule 404(b) because the defense did not offer it as character/propensity proof; it was offered to corroborate the defendant’s state-of-mind narrative.
• The error was not harmless under Rule 52(a) given that intent was central, the excluded evidence was not merely cumulative, and the government’s proof on intent was not overwhelming.

3) Analysis

A. Precedents Cited

• United States v. Desposito, 704 F.3d 221 (2d Cir. 2013) (standard of review for evidentiary rulings).
  Influence: The panel framed review as deferential but not toothless, setting up the inquiry whether the district court’s ruling rested on an erroneous legal view.
• United States v. Barret, 848 F.3d 524 (2d Cir. 2017) (abuse-of-discretion definition: erroneous view of law/clearly erroneous assessment/outside permissible range).
  Influence: Provided the doctrinal hook for concluding exclusion was outside the permissible range because it incorrectly treated corroborative evidence as propensity evidence.
• United States v. Mercado, 573 F.3d 138 (2d Cir. 2009) (evidentiary errors are subject to harmless error analysis).
  Influence: Ensured the court separately evaluated prejudice and did not treat error as automatically reversible.
• United States v. Anderson, 747 F.3d 51 (2d Cir. 2014) (conspiracy requires proof of specific intent to commit the object offenses).
  Influence: Central to the court’s insistence that if the jury credited Roldan’s belief he was facilitating a seizure (not importation), he lacked the intent required for conviction even if he acted “improperly, unethically, or even illegally” in some other way.
• United States v. Detrich, 865 F.2d 17 (2d Cir. 1988) (corroborative evidence that supports defendant’s version of key conversations may be admissible and its exclusion may be non-harmless where intent is central).
  Influence: This was the court’s closest analogue. As in Detrich, the relevance lay in corroborating what the defendant said he was told (there: wedding plans; here: Aguas’s seizure-and-reward practice), which in turn made the defendant’s innocent-explanation narrative more credible on intent. The panel emphasized the inferential chain and found it even stronger here because the prior seizures were undisputed.
• United States v. Al Kassar, 660 F.3d 108 (2d Cir. 2011) (affirming exclusion where defendant offered prior lawful conduct to suggest lawful conduct on the charged occasion).
  Influence: The government relied on Al Kassar for a “good acts” propensity bar; the panel distinguished it as inapposite because Roldan’s theory did not ask the jury to infer Aguas acted “in conformity” with past good conduct, but rather to credit Roldan’s testimony about what Aguas told him and what Roldan believed.
• United States v. Figueroa, 548 F.3d 222 (2d Cir. 2008) (appellate court should not affirm on Rule 403 grounds where the district court did not exercise discretion or conduct the balancing).
  Influence: The government argued Rule 403 on appeal, but the panel refused to affirm on that basis because the district court did not conduct a Rule 403 analysis and the government’s pretrial motion relied solely on Rule 404(b).
• United States v. Dukagjini, 326 F.3d 45 (2d Cir. 2003) (harmlessness standard: “highly probable” the error did not affect the verdict).
  Influence: Provided the operative prejudice test used to reject harmlessness.
• Kotteakos v. United States, 328 U.S. 750 (1946) (reversal required unless the court is sure error did not influence jury or had but very slight effect).
  Influence: Reinforced the strictness of the harmless-error inquiry in close evidentiary/credibility cases.
• United States v. Litvak, 808 F.3d 160 (2d Cir. 2015) and United States v. Oluwanisola, 605 F.3d 124 (2d Cir. 2010) (harmless-error factors: importance, cumulativeness, overall strength).
  Influence: Structured the panel’s application of harmlessness to the record, focusing on whether the excluded evidence was central and distinct rather than duplicative.
• United States v. McCallum, 584 F.3d 471 (2d Cir. 2009) (strength of government’s case as “most critical factor”).
  Influence: Supported the panel’s conclusion that the government’s proof on intent was not overwhelming (particularly given Roldan’s low-level role and lack of demonstrated capability at the airport).
• United States v. Forrester, 60 F.3d 52 (2d Cir. 1995) and United States v. Blum, 62 F.3d 63 (2d Cir. 1995) (errors “going to the heart” of a critical issue are less likely harmless).
  Influence: Justified heightened sensitivity to exclusion of evidence bearing on the core issue—intent—where the defense rose or fell on credibility.
• United States v. Zhong, 26 F.4th 536 (2d Cir. 2022) (government bears burden to prove evidentiary error harmless).
  Influence: The panel explicitly placed the burden on the government and found it unmet.
• United States v. Aboumoussallem, 726 F.2d 906 (2d Cir. 1984) (third-party similar-acts evidence offered by defendant generally lacks prejudice; key question is relevance).
  Influence: Discussed in a footnote as an alternative admissibility theory—third-party acts to show third party’s plan/motive/intent—though the court did not need to reach it because it found admissibility on corroboration/state-of-mind grounds.

B. Legal Reasoning

1) Relevance (Rule 401): The court treated Roldan’s intent as the decisive battleground. The excluded evidence—Major Torres’s statement that Aguas twice previously provided tips leading to successful seizures—made it more likely that Aguas told Roldan the story Roldan recounted (that Aguas had a reward-based practice of supplying seizure tips through anti-narcotics channels). That corroboration, in turn, made Roldan’s claimed belief more plausible. The court emphasized that the evidence directly affected the jury’s evaluation of Roldan’s credibility on the “why” of his admitted communications with the purported trafficker.

2) Rule 404(b) mischaracterization: The key doctrinal move is the panel’s separation of: (a) propensity use (“Aguas previously did good law-enforcement work; therefore he did so here”), from (b) corroboration/state-of-mind use (“Aguas previously had seizure-tip experience; therefore it is more likely he described that experience to Roldan; therefore it is more likely Roldan believed the ‘seizure’ narrative and lacked criminal intent”). On the panel’s account, the defense theory did not require the jury to decide what Aguas “really intended” in the charged episode; it required the jury to decide what Roldan believed and intended. Because the proffered evidence served that latter function, it was not excluded by Rule 404(b)(1).

3) Rule 403 not a permissible appellate salvage: The government attempted to justify exclusion as unfairly prejudicial/confusing under Rule 403, but the panel declined because the district court did not conduct a Rule 403 balancing and the government did not present that basis below, invoking United States v. Figueroa.

4) Non-harmlessness (Rule 52(a)): The panel found prejudice for three interlocking reasons:

• Centrality: Intent was the main disputed issue; the evidence went to the heart of the defense by corroborating the conversation that supposedly formed Roldan’s belief.
• Not cumulative: Although the jury heard that Aguas reported a tip to Major Torres (and met with the DEA), that evidence also contained damaging context: the DEA declined to work with the source and Torres instructed Aguas not to act independently—facts the government used to argue there was no authorized operation. The excluded “prior seizures” evidence performed a distinct function: it supported a defense theory that Roldan believed in a seizure-for-reward scheme even if it was not a formally authorized undercover operation.
• Government’s intent case not overwhelming: The court noted the absence of proof that Roldan had airport connections, experience, or the capability to move massive quantities through Cartagena, and recognized that certain conduct (e.g., emphasis on warehousing) could be consistent with a seizure plan.

C. Impact

1) Practical evidentiary impact in intent-centered cases: The decision strengthens defendants’ ability—particularly in conspiracy cases where intent is inferred from communications—to introduce third-party “other acts” evidence when the purpose is to corroborate a state-of-mind account (what the defendant was told and therefore believed), rather than to prove anyone’s propensity.

2) Narrowing overbroad Rule 404(b) objections: The opinion cautions district courts against reflexively labeling third-party background conduct as impermissible character evidence where the real relevance is corroboration of a conversation central to the defendant’s mens rea defense.

3) Appellate discipline on Rule 403: By refusing to affirm on Rule 403 absent a district court balancing, the panel reinforces that trial courts should make explicit 403 determinations when excluding defense evidence that goes to a core element like intent—both to guide the trial and to permit meaningful appellate review.

4) Harmless-error rigor when credibility is pivotal: Where the excluded evidence would have supplied non-cumulative corroboration of the defendant’s narrative on intent, the Second Circuit signaled skepticism toward harmlessness arguments, especially where the government exploits the resulting “evidentiary gap” in summation.

4) Complex Concepts Simplified

• Specific intent (in conspiracy): The government must prove the defendant meant to join the illegal plan and meant to further its unlawful objective. Here, if Roldan truly believed he was helping stage a seizure (not importation), he would lack the specific intent to import cocaine even if his behavior was otherwise improper.
• Relevance (Rule 401): Evidence is relevant if it makes a consequential fact even slightly more or less likely. Corroboration of a key conversation can be relevant because it affects credibility and thus the likelihood the defendant held the claimed belief.
• Rule 404(b): This rule blocks “other acts” evidence when used to argue “he’s the kind of person who would do it again.” It does not block other-acts evidence used for a different logical purpose (here, corroborating what was said to the defendant and therefore what the defendant believed).
• Rule 403: Even relevant evidence can be excluded if its probative value is substantially outweighed by dangers like unfair prejudice or confusion. But appellate courts are reluctant to uphold exclusion on 403 grounds when the trial court did not actually weigh those factors on the record.
• Harmless error: Not every mistake triggers reversal. But when excluded evidence bears directly on the central disputed issue (intent) and could have influenced the jury’s credibility assessment, the government must show it is highly probable the verdict was unaffected—an especially difficult showing in close, narrative-driven cases.

5) Conclusion

United States v. Cardenas underscores that mens rea defenses often rise or fall on whether jurors believe the defendant’s account of what he was told and why he acted. The Second Circuit held that evidence of a third party’s prior similar conduct—undisputed here—may be admissible when it corroborates the defendant’s testimony about those pivotal conversations and thus the defendant’s state of mind. Treating such corroborative proof as barred propensity evidence under Rule 404(b) is legal error, and where intent is central and the proof is not cumulative, the error may require vacatur and remand.