Credible Innocence Showing Required to Withdraw a Guilty Plea; Bare National Statistics Do Not Establish Unwarranted Sentencing Disparity

1. Introduction

In United States v. Miller (10th Cir. June 17, 2026), the Tenth Circuit affirmed (1) the denial of Gregory Miller’s motion to withdraw his guilty plea and (2) his 360-month sentence for production of child pornography under 18 U.S.C. § 2251(a) and (e). Mr. Miller had pleaded guilty without a plea agreement after a Rule 11 plea colloquy before a magistrate judge. After the PSR issued, he sought to withdraw his plea, asserting he “did not do” the charged conduct and arguing that personal stressors and untreated ADHD affected his decision-making. He also challenged his sentence as procedurally and substantively unreasonable, emphasizing national sentencing statistics for similarly scored defendants.

The appeal raised two recurring issues in federal criminal practice: (a) what it takes to show a “fair and just reason” to withdraw a guilty plea under Fed. R. Crim. P. 11(d)(2)(B), and (b) what constitutes an adequate explanation—and adequate disparity analysis—when imposing a within-Guidelines sentence at the statutory maximum.

2. Summary of the Opinion

The court affirmed across the board. On plea withdrawal, it held the district court did not abuse its discretion in concluding that three core Gordon factors weighed against Mr. Miller: (1) his asserted innocence was not credible given his prior admissions during the plea process; (5) the ineffective-assistance claim was not reviewable on direct appeal because the record was underdeveloped; and (6) the plea was knowing and voluntary as reflected in the Rule 11 colloquy and Mr. Miller’s sworn confirmations.

On sentencing, the panel concluded the district court adequately explained the sentence under 18 U.S.C. § 3553(a) and permissibly discounted broad national statistics offered to show disparity because they lacked “specific comparators” and did not capture individualized circumstances. Because the 360-month sentence was within the Guidelines (as limited by the statutory maximum), it enjoyed a rebuttable presumption of reasonableness that Mr. Miller did not overcome.

3. Analysis

3.1 Precedents Cited

  • United States v. Gordon, 4 F.3d 1567 (10th Cir. 1993): Source of the seven-factor framework for evaluating whether a defendant has shown a “fair and just reason” to withdraw a plea.
  • United States v. Hamilton, 510 F.3d 1209 (10th Cir. 2007): Quoted for the Gordon factors and the rule that if factors (1) assertion of innocence, (5) assistance of counsel, and (6) knowing/voluntary plea weigh against the defendant, the court need not address the remaining factors.
  • United States v. Marceleno, 819 F.3d 1267 (10th Cir. 2016): Provided standards of review (abuse of discretion generally; de novo on knowing/voluntary plea), reiterated that an innocence assertion must be credible, and explained that some “colorable basis” must support the claim.
  • United States v. Carr, 80 F.3d 413 (10th Cir. 1996): Cited for the statement that motions to withdraw pre-sentencing are treated liberally and (in tension with other cases) for language suggesting a defendant need only “assert his innocence.”
  • United States v. Hickok, 907 F.2d 983 (10th Cir. 1990): Treated as the earlier anchor for the proposition that a mere subjective assertion of innocence does not compel plea withdrawal; used to support the “credibility” requirement.
  • United States v. Byrum, 567 F.3d 1255 (10th Cir. 2009): Reinforced the credibility requirement for asserted innocence and the need to address counsel-assistance as part of the plea-withdrawal analysis.
  • United States v. Suggs, 998 F.3d 1125 (10th Cir. 2021): Cited for the intra-circuit conflict principle—when decisions conflict, the earliest decision controls—supporting the panel’s choice to follow Hickok over Carr to the extent of tension.
  • United States v. Kearn, 90 F.4th 1301 (10th Cir. 2024): Cited for appellate deference to district-court credibility findings, important where the withdrawal motion turns on believability.
  • United States v. Sanchez-Leon, 764 F.3d 1248 (10th Cir. 2014): Provided two key rules—(1) ineffective-assistance claims are presumptively dismissible on direct appeal absent a fully developed record, and (2) sentencing courts imposing within-Guidelines sentences need only provide a general statement of reasons and need not respond to every leniency argument.
  • United States v. Kramer, 168 F.3d 1196 (10th Cir. 1999): Cited in support of relying on the plea colloquy record to reject claims that a plea was not knowing and voluntary.
  • United States v. Vazquez-Garcia, 130 F.4th 891 (10th Cir. 2025): Cited for the two-step reasonableness framework—procedural reasonableness then substantive reasonableness.
  • United States v. Rocha, 145 F.4th 1247 (10th Cir. 2025): Used for the abuse-of-discretion standard on sentencing and the requirement that the district court explain the sentence in sufficient detail to permit appellate review.
  • United States v. Sandoval, 959 F.3d 1243 (10th Cir. 2020): Cited for the rebuttable presumption of reasonableness for within-Guidelines sentences.
  • United States v. Lucero, 130 F.4th 877 (10th Cir. 2025): Used both for the presumption framework and for the proposition that national statistics are not dispositive on disparity because they do not capture individualized history/characteristics.
  • United States v. Guevara-Lopez, 147 F.4th 1174 (10th Cir. 2025): Cited for the rule that “bare national statistics without specific comparators” cannot defeat an otherwise sufficient sentencing explanation.
  • United States v. Grigsby, 749 F.3d 908 (10th Cir. 2014), and United States v. Adams, No. 22- 2071, 2023 WL 3266822 (10th Cir. May 5, 2023): Foreclosed the policy-based attack on U.S.S.G. § 2G2.1 as allegedly not “empirical.”
  • United States v. Wilcox, No. 20-7047, 2022 WL 2253948 (10th Cir. June 23, 2022): Used by analogy to uphold rejecting an innocence claim as not credible in light of prior plea-hearing admissions.

3.2 Legal Reasoning

A. Plea Withdrawal: The Credibility Requirement for “Assertion of Innocence”

The opinion’s central clarification is methodological: when evaluating the first Gordon factor, an assertion of innocence must be credible, not merely stated. Mr. Miller urged the court to follow the phrasing in United States v. Carr suggesting that a defendant need only “assert his innocence.” The panel rejected that view by anchoring the standard in United States v. Hickok (earlier in time) and the post-Carr line of cases (United States v. Marceleno, United States v. Byrum, United States v. Hamilton) that explicitly require credibility and a “colorable basis.”

To resolve any tension, the panel invoked the intra-circuit rule highlighted in United States v. Suggs: where there is conflict, the earliest controlling decision prevails. Because Hickok predates Carr, and because later cases harmonize with Hickok, the district court correctly evaluated whether Mr. Miller’s assertion of innocence was believable in light of the full record.

Applying that standard, the district court permissibly found Mr. Miller’s claim not credible because it was contradicted by his sworn plea petition and Rule 11 admissions. The panel emphasized deference to credibility findings under United States v. Kearn, and it rejected the argument that appellate judges were “in the same position” as the district court merely because the plea occurred before a magistrate judge; the district judge assessed Mr. Miller’s credibility directly at the withdrawal hearing.

B. Ineffective Assistance of Counsel: Direct Appeal Limitations

Mr. Miller attempted to use the fifth Gordon factor—“defendant’s assistance of counsel”—to support withdrawal, while also acknowledging the record was not developed enough for an ineffective-assistance ruling on direct appeal. The panel applied United States v. Sanchez-Leon: ineffective-assistance claims are “presumptively dismissible” on direct appeal unless fully developed. It therefore declined to reach the merits, treating the district court’s brief discussion as a practical necessity to address the Gordon framework rather than an impermissible merits adjudication.

C. Knowing and Voluntary Plea: Rule 11 Record Controls

Mr. Miller argued that divorce papers served shortly before the plea hearing and untreated ADHD meant he lacked the requisite understanding. The panel affirmed the district court’s reliance on the Rule 11 record: Mr. Miller repeatedly affirmed soundness of mind, an understanding of consequences, and culpability. Citing United States v. Kramer and United States v. Marceleno, the court treated those sworn statements as strong evidence the plea was knowing and voluntary, absent concrete contrary indicators in the record.

D. Consequence of the “Three-Factor” Finding

Under United States v. Hamilton, if factors (1), (5), and (6) weigh against the defendant, the district court need not analyze the remaining factors (prejudice to the government, delay, inconvenience to the court, and waste of judicial resources). The panel held that approach was proper here, which functionally narrows many plea-withdrawal disputes: once credibility, counsel-assistance posture, and voluntariness are resolved against the defendant, the motion often fails without extended factor-by-factor balancing.

E. Sentencing: Explanation, Disparity, and the Limits of National Statistics

On sentencing, the court applied the familiar two-step inquiry from United States v. Vazquez-Garcia (procedural then substantive reasonableness) and the deferential abuse-of-discretion standard described in United States v. Rocha. With a total offense level of 42 and criminal history category I, the Guidelines would have been 30 years to life, but the statutory maximum capped the sentence at 30 years under 18 U.S.C. § 2251(e) and U.S.S.G. § 5G1.1(c)(1).

Mr. Miller’s principal procedural complaint was that the district court’s § 3553(a) discussion was “conclusory,” and that the court did not meaningfully address § 3553(a)(6) disparity concerns raised by national statistics (average and median sentences under U.S.S.G. § 2G2.1 for CHC I defendants).

The panel rejected those arguments by combining three doctrinal points:

  • Under United States v. Sanchez-Leon, when imposing a within-Guidelines sentence, a district court needs only a general statement of reasons and need not address every leniency argument.
  • Under United States v. Lucero, national statistics are not dispositive because they lack individualized context.
  • Under United States v. Guevara-Lopez, “bare national statistics without specific comparators” cannot undermine an otherwise adequate explanation.

Substantively, because the sentence was within the Guidelines (as capped), it carried a presumption of reasonableness under United States v. Sandoval, and Mr. Miller did not rebut it. The panel also noted that his policy-based critique of U.S.S.G. § 2G2.1 was foreclosed by United States v. Grigsby (and reaffirmed by United States v. Adams).

3.3 Impact

Although issued as a nonprecedential “Order and Judgment,” the opinion consolidates and crisply applies two practical rules that are likely to shape litigation behavior in the circuit:

  • Plea withdrawal: defendants should expect that an “assertion of innocence” will be tested for credibility against the Rule 11 record; a post-plea denial that conflicts with sworn admissions will rarely carry the first Gordon factor, especially where the district court makes an adverse credibility finding.
  • Sentencing disparity arguments: generalized national averages—without identified “specific comparators” or individualized parallels—are unlikely to compel a different sentence or a more elaborate district-court explanation, particularly for within-Guidelines sentences.

The opinion also underscores a strategic reality for defense counsel: ineffective-assistance contentions typically will not provide a direct-appeal vehicle for undoing a plea; they are more commonly reserved for collateral review where the record can be developed.

4. Complex Concepts Simplified

  • Rule 11 plea colloquy: the judge’s on-the-record questioning to ensure the defendant understands the charge, rights waived, potential penalties, and that the plea is voluntary. Sworn answers here are powerful evidence against later claims of confusion.
  • “Fair and just reason” (Rule 11(d)(2)(B)): the legal threshold for withdrawing a plea before sentencing; it is not automatic and is assessed using the Gordon factors.
  • Gordon factors: seven considerations guiding plea-withdrawal decisions, including innocence, counsel, voluntariness, timing, prejudice, inconvenience, and judicial resources.
  • Credible assertion of innocence: more than saying “I’m innocent”; the court looks for a plausible, supported basis and consistency with prior sworn admissions.
  • Procedural vs. substantive reasonableness: “procedural” asks whether the court followed correct steps (Guidelines calculation and explanation); “substantive” asks whether the length of the sentence is reasonable in light of § 3553(a).
  • Within-Guidelines presumption: on appeal, a sentence within the advisory Guidelines range is presumed reasonable unless the defendant shows why it is not.
  • § 3553(a)(6) sentencing disparities: the goal of avoiding “unwarranted” disparities; broad statistics may be less persuasive than comparisons to genuinely similar defendants and offenses.

5. Conclusion

United States v. Miller reinforces that plea withdrawal in the Tenth Circuit turns heavily on the integrity of the Rule 11 record and the credibility of post-plea claims—particularly the claimed “assertion of innocence,” which must be colorably supported and consistent with the broader record under United States v. Hickok and United States v. Marceleno. On sentencing, it confirms that within-Guidelines sentences require only a general explanation and that generalized national statistics, without specific comparators, typically will not establish unwarranted disparity under United States v. Lucero and United States v. Guevara-Lopez. The decision thus serves as a practical blueprint: sworn plea admissions are difficult to unwind, and disparity arguments are most effective when grounded in truly comparable cases rather than aggregate averages.