This article originally was published by the Federal Sentencing Reporter, University of California Press, vol. 35, issue 2 (Dec. 2022).


Twenty-seven years ago, I wrote for this publication addressing a few places that I thought the Federal Sentencing Guidelines (“guidelines”) needed to be tweaked. In the present Article, I will touch on those recommendations again and assess the current state of the now discretionary guidelines. First, I should admit that my current views may have moved more toward the defense side of things, given that it has been a few years since I was with the Department of Justice (DOJ). However, I would like to think that my current posture is built around an appreciation for a sentencing system that is still very much, in my view, a work in progress. It is also worth noting that it has been thirty-three years since the statutory framework creating the guidelines was upheld by the Supreme Court in its 1989 opinion in Mistretta v. Ohio. At the time, I was a Deputy Assistant Attorney General in the Criminal Division of Main Justice. I recall being staffed by my boss, Assistant Attorney General William F. Weld, to lead the effort to put together a handbook for DOJ lawyers at Main Justice and assistant U.S. Attorneys in the ninety-four federal judicial districts on the guidelines and their proper interpretation by federal prosecutors. Most of the credit for the effort must go to Mark Robinson, then Special Counsel to Weld. Mark enlisted a small cadre of Criminal Division attorneys to help generate what became known as the “Red Book.” This booklet was the first attempt by the DOJ to educate prosecutors on this new approach to sentencing mandated by Congress.

Almost all roads lead to sentencing in federal criminal cases. In fiscal year 2021, the vast majority of offenders (98.3%) pleaded guilty. This rate has been consistent for more than twenty years but reached an all-time high last year.1 The DOJ’s strategy historically, pre- and postguidelines, has always been to create an environment that makes the trial of any criminal case a very risky proposition for defendants and defense counsel in terms of sentencing length. Once an indictment is returned or a complaint is filed, the die may be cast for many defendants under guideline sentencing. The few of us still practicing who remember the sentencing system before guidelines recall the horse-trading atmosphere created by the different sentencing perspectives held by each lifetime-appointed district court judge across the United States. The process was much more art than science.

Although the guidelines are now advisory, not mandatory, they still provide important guidance for district courts today.2 Sentencing has become more akin to a science, requiring much more analysis by prosecutors, defense counsel, probation officers, and judges—who can now consult, in a broad spectrum of cases, a sentencing “industry” of advisors and experts on the most up-to-date factors and precedent that might result in a departure or variance for a particular defendant. At the same time, prosecutorial discretion is as broad as ever, and a particular prosecutor’s perspective remains a central factor in a defendant’s balancing of considerations. Despite official policy stating that “once the decision to prosecute has been made, the attorney for the government should charge and pursue the most serious, readily provable offenses,” and “[i]t is important to the fair and efficient administration of justice in the federal system that the government bring as few charges as are necessary to ensure that justice is done,” these tenets are not always followed in practice.3 For example, the aggravated identity theft offense under 28 U.S.C. § 1028A carries a two-year mandatory minimum prison sentence that can be tacked on to more than sixty predicate federal offenses.4 Today, prosecutors can attach this mandatory minimum offense to as many charges as they wish in an indictment, which gives them enormous leverage in reaching a plea deal with a defendant looking at potentially years of stacked sentences. Thus, in addition to the broad judicial discretion under the now advisory guidelines, significant uncertainty and subjectivity exist in the process long before the sentencing hearing.

Coming Full Circle

In some ways, we have come full circle from 1975, when Senator Edward Kennedy first introduced a bill to address disparities in sentencing across the United States. This legislation was introduced, and ultimately adopted, because substantive data supported the growing concern about disparities in sentencing. Among many major changes, the Sentencing Reform Act of 1984 created the U.S. Sentencing Commission (“Commission”), which was directed to develop an appropriate sentencing range for each category of offense involving each category of defendant.5 Specifically, the Commission’s charge was to “provide certainty and fairness in meeting the purposes of sentencing, avoiding unwarranted sentencing disparities among defendants with similar records who have been found guilty of similar criminal conduct, while maintaining sufficient flexibility to permit individualized sentences when warranted by mitigating or aggravating factors not taken into account in the establishment of general sentencing practices.”6 Indeed, in fashioning an appropriate sentence, judges are directed by statute to consider “the need to avoid unwarranted sentencing disparities among defendants with similar records who have been found guilty of similar conduct.”7 Although this provision is “primarily aimed at national disparities, rather than those between codefendants,”8 courts have recognized that a sentence may be “substantively unreasonable because of the disparity with the sentence given to a codefendant.”9 The Commission was also instructed to “develop means of measuring the degree to which the sentencing, penal, and correctional practices are effective in meeting the purposes of sentencing.”10

The Sentencing Reform Act was, of course, subject to challenges and underwent certain adjustments in practical application before the Supreme Court, but they remained mandatory until the Court’s 2005 decision in United States v. Booker.11 In Booker, the Court excised the provisions of the guidelines that made them mandatory in application, stating that “district courts, while not bound to apply the Guidelines, must consult those Guidelines and take them into account when sentencing.” In so holding, the Court acknowledged that this was “not the system Congress enacted,” but it “nonetheless continue[d] to move sentencing in Congress’ preferred direction.” Since then, the guidelines have been advisory in nature, but they remain the starting point for all federal sentencing: a district judge must calculate a criminal history level and offense level and use that information to determine the advisory guideline range. Only then can the judge adjust—upward or downward—on the basis of relevant § 3553(a) factors. Thereafter, on review, an appellate court must “ensure that the district court committed no significant procedural error, such as failing to calculate (or improperly calculating) the Guidelines range, treating the Guidelines as mandatory, [or] failing to consider the § 3553(a) factors.”12

This is the federal sentencing framework we live with today, but it must be asked whether striking the mandatory provisions of the guidelines “continued to move sentencing in Congress’ preferred direction.” Unlike in 1975—or 1984, or even 2005—we now have technology at our fingertips that allows us to amass huge amounts of data to better analyze sentencing disparities and other statistical evidence of the real-life consequences of the guidelines. For better or worse, we can now see with amazing clarity the disparities in sentencing that exist across federal circuits and districts, depending on offense type, demographic profile, time period, and criminal history.

The Data

In its January 2020 report on “Inter-District Differences in Federal Sentencing Practices,” the Commission stated that variations in sentencing practices across districts have increased since the Booker decision in 2005, and since the Commission’s last report and analysis published in 2012.13 The Commission also stated that “certain districts have consistently sentenced more—or less—severely in relation to the guideline minimums than other districts, both over time and across offense type.” For example, the Northern District of Texas, Northern District of Iowa, Southern District of Mississippi, and Middle District of North Carolina appear among the ten highest-severity districts for §§ 2B1.1 (theft offenses), 2D1.1 (unlawful manufacturing, importing, exporting, or trafficking), 2K2.1 (receipt, possession, or transport of firearms or ammunition), and 2L1.2 (unlawfully entering or remaining in the United States). Several other districts are consistent in at least three primary guidelines—the Southern District of Georgia appears in all but § 2L1.2, and the Eastern District of North Carolina and Eastern District of Texas in all but § 2B1.1. Other districts consistently appear among those farthest below the average for multiple guidelines in the Post-Report Period. In fact, four appear for all but § 2L1.2 (District of Rhode Island; District of Utah; Eastern District of Wisconsin; Southern District of Ohio), and one (District of Oregon) appears for each of the four guidelines.

Despite the Commission’s not-so-subtle coding of these results in red for the top ten and blue for the bottom ten districts, it is also quick to list a number of factors likely to contribute to sentencing disparity, including caseload volume, variation in prosecutorial practices, and the fact that certain kinds of cases are disproportionately more or less frequent in certain districts than in others. But, perhaps more importantly than the why, the data clearly show that sentencing disparity is far from random—there are consistent, steady disparities in sentencing across states, across circuits, and even among districts within the same state.

The disparities do not stop there. In its most recent 2017 report on “Demographic Differences in Sentencing,” the Commission stated that Black male offenders received sentences that were, on average, 19.1% longer than those of similarly situated White male offenders between 2012 and 2016.14 The Commission further concluded that nongovernment-sponsored departures and variances appeared to contribute significantly to the difference in sentence length between Black male and White male offenders. Female offenders of all races, however, received shorter sentences than White male offenders. Of course, it may be that the actual range of disparity has nevertheless improved overall since the introduction of the guidelines, when this kind of data was not available. Nevertheless, we must hope that these public resources continue to spur legislative, prosecutorial, or judicial rethinking of the import and implications of the guidelines. The simple fact remains that we can, and should, do more to address sentencing disparity among similarly situated defendants.

Reconsidering My Recommendations

I made several recommendations in my 1995 article. First, I recommended that courts should “be able to decide whether defendants are to receive a reward for providing substantial assistance—as they could before the guidelines.” However, prosecutors still control whether a motion is filed under USSG 5K1.1, and the court can consider a reduction based on substantial assistance only upon a prosecutor’s motion. A district judge can go below a statutory mandatory minimum sentence only upon motion by the government.15 Despite this gatekeeping function held by prosecutors, there is no doubt that substantial assistance is a major component in federal sentencing today. In 2021, 12.7% of downward departures in sentencing occurred by reason of substantial assistance provided by the defendant, upon government motion.16 This happened most frequently in drug trafficking cases, but it occurred in a variety of other kinds of cases as well.

Second, I recommended that the guidelines be revised to send a “clear message to judges that they should take the personal characteristics of the defendant into account in determining a sentence.” I noted that the policy statements in § 5H concerning the relevance of offender characteristics were internally inconsistent with Congress’s instruction to judges to consider the § 3553(a) factors. Since 1995, this section has been amended several times, and the introductory commentary specifically states that these policy statements are “evolutionary in nature.”17 As has been true since 1987, factors of race, sex, national origin, creed, religion, and socioeconomic status are not relevant in considering a departure. However, in certain circumstances, a departure may be based on a defendant’s mental and emotional condition or on the loss of caretaking/financial support provided to their family.

Third, I opined that the provisions on relevant conduct in USSG 1B1.3 gave judges too much information to consider in sentencing, including considering conduct for which the jury has acquitted the defendant, or evidence that was suppressed at trial under the Fourth Amendment. I noted that many state sentencing commissions had diverged from these rules, and I suggested that the Commission do the same. Despite many amendments over the years, the basic rule remains unchanged. The guidelines state, as of 2018, that “[c]onduct that is not formally charged or is not an element of the offense of conviction may enter into the determination of the applicable guideline sentencing range. The range of information that may be considered at sentencing is broader than the range of information upon which the applicable sentencing range is determined.”18

In revisiting these recommendations I made twenty-seven years ago, I am struck that the framework of federal sentencing is as complex and uncertain as ever, perhaps even more so now. The main difference, it seems, is that defendants and their counsel are less certain about the sentencing results likely to be handed down—not because the guidelines are unclear, but because the defendant’s location and demographics must be considered. For example, a defendant in Georgia convicted at trial of participating in a fraudulent scheme valued at $17 million was sentenced to fourteen years in prison in 2022, whereas certain defendants convicted at trial of participating in a $63 million fraud scheme in Florida received five- to six-year sentences in 2018.19 Perhaps there were other factors that warranted the disparity between these cases in terms of imprisonment time and loss amounts, yielding dramatically different sentences within a single circuit, but the happenstance of location cannot be a barometer of just results. Even though some defendants may receive more lenient sentences under the advisory guidelines framework, this has not resulted in further progress toward the overall goal of fundamental fairness in our justice system.

I ended my 1995 article with the hope that “the new Commission will learn from the guidelines’ first seven years to improve them for the future.” I can now note that, as of August 2022, the Sentencing Commission has filled several vacancies and has a voting quorum for the first time in more than three years.20 I maintain the hope that, with decades’ more experience and the vast amounts of data on federal sentencing now available, the Commission will continue to revise and improve the impact of the guidelines for defendants across the nation.

* Thank you to Luke Cass (Partner) and Allison Pearl (Associate) of Womble Bond Dickinson (US) LLP for their assistance in writing this article.
1 U.S. Sentencing Comm’n, Overview of Federal Criminal Cases: Fiscal Year 2021 8 (Apr. 2022),
2See Kimbrough v. United States, 552 U.S. 85, 107 (2007) (recognizing that the Sentencing Commission plays a “key role” and “fills an important institutional role”); Gall v. United States, 552 U.S. 38, 49 (2007) (“As a matter of administration
and to secure nationwide consistency, the Guidelines should be the starting point and the initial benchmark.”).
3Id., citing USAM 9-27.000—Principles of Federal Prosecution,
4 Congressional Research Service, Mandatory Minimum Sentencing:
Federal Aggravated Identity Theft (July 15, 2022),
5 28 U.S.C. § 991.
6 28 U.S.C. § 991(b)(1)(B).
7 18 U.S.C. § 3553(a)(6).
8 United States v. Rivera–Gonzalez, 626 F.3d 639, 648 (1st Cir. 2010).
9 United States v. Correa–Osorio, 784 F.3d 11, 28 n. 25 (1st Cir. 2015) (“[S]entencers can consider disparities between codefendants, we have noted—even though § 3553(a)(6) chiefly addresses disparities among defendants nationwide.”); United States v. Martin, 520 F.3d 87, 94 (1st Cir. 2008) (“[D]istrict courts have discretion, in appropriate cases, to align codefendants’ sentences somewhat in order to reflect
comparable degrees of culpability—at least in those cases where disparities are conspicuous and threaten to undermine confidence in the criminal justice system.”).
10 28 U.S.C. § 991(b)(2).
11 543 U.S. 220 (2005).
12Gall, 552 U.S. at 38.
13 U.S. Sentencing Comm’n, Inter-District Differences in Federal Sentencing Practices: Sentencing Practices Across Districts from 2005–2017,
14 U.S. Sentencing Comm’n, Demographic Differences in Sentencing: An Update to the 2012 Booker Report (Nov. 2017),
15 18 U.S.C. § 3553(e).
16 U.S. Sentencing Comm’n, 2021 Annual Report and Sourcebook of Federal Sentencing Statistics, https://www.ussc.
17 U.S. Sentencing Comm’n, Guidelines Manual 2021, 2018 Chapter 5,
18 U.S. Sentencing Comm’n, Guidelines Manual 2021, 2010 Chapter 1,
19 Rosie Manins, Ex-Atlanta Exec Jailed 14 Years in $17 M Bribery Scheme, Law360 (Sept. 8, 2022),; U.S. v. Crabtree, 878 F.3d 1274 (2018).