In late December, the Department of Justice's Antitrust Division (“Division”) announced it would not appeal a federal district court's dismissal of the indictment in U.S. v. Stolt-Nielson S.A., stating that “it respects the role of the court in making the factual determination” that the amnesty agreement the government had entered into with the defendants had not been breached. The government had revoked the amnesty agreement and sought the indictment based on its belief that the defendants breached the agreement by failing to cease their anticompetitive conduct immediately upon uncovering it. This case represented the first time that the Antitrust Division sought to remove a company from its Corporate Leniency Program since that program was adopted in 1978.

The Division's announcement followed the November 30, 2007 decision by the United States District Court for the Eastern District of Pennsylvania dismissing the indictment. The Court held that Stolt-Nielson had not breached the amnesty agreement and had cooperated with the government giving it the benefit of the bargain; therefore, the government’s conduct in unilaterally revoking the agreement and obtaining indictments was “fundamentally unfair."

Criminal Antitrust Investigations and Prosecution and the Corporate Leniency Program

A criminal antitrust investigation routinely lasts three to five years and the targets of investigations face penalties that include restitution to victims, jail time for offenders, and corporate criminal fines that often exceed $100 million. In addition, criminal investigations frequently spawn other legal actions, including civil suits seeking treble damages, antitrust prosecution by foreign jurisdictions, and even state antitrust prosecutions. As a consequence, once corporate counsel discovers evidence of an antitrust violation, all options that limit the liability of the corporation and its employees must be weighed, including an important, but potentially overlooked, option – the Antitrust Division’s Corporate Leniency Program (“amnesty program”).

Over the past ten years, the prosecution of antitrust violations has changed significantly. The combination of stronger antitrust enforcement, increased fines and jail terms, and the amnesty program has contributed to a sharp increase in criminal antitrust prosecutions. Criminal antitrust activities are inherently secretive and are therefore difficult to detect. In recent years, however, the Antitrust Division has added tools beyond the use of investigative grand juries and cooperating informants, including its aggressive use of search warrants, consensual monitoring with sophisticated audio and video surveillance equipment, cooperation with foreign authorities, and its expansion of the amnesty program to obtain evidence from co-conspirators about conspiracies.

The Antitrust Division created the Corporate Leniency Program in 1978 and revised it in 1993 to facilitate the discovery and prosecution of antitrust violations. Under the program, the first member of a cartel to self-report certain anticompetitive conduct can receive protection from prosecution, including complete amnesty from criminal convictions for the firm, and for its directors, officers and employees, including the avoidance of any fines and jail sentences. Such benefits accrue only if the Antitrust Division has not already received information about the illegal activity, and the following other conditions are met:

  • The amnesty applicant took prompt and effective action to cease the activity upon its discovery;
  • The applicant’s cooperation is candid and complete and continuing and is truly a corporate act;
  • The applicant makes restitution to the victims (where possible); and
  • The applicant was not the leader or originator of the illegal activity and did not coerce another party.
  • While it is possible to receive amnesty after the Antitrust Division has opened an investigation, the chances of obtaining a grant of amnesty are exceedingly low
  • where the investigation has already yielded significant evidence of a criminal violation.


Stolt-Nielsen sought entry into the Corporate Leniency Program after its chief executive officer learned of its participation in a customer allocation conspiracy with two of its competitors in the parcel tanker shipping industry, Odfjell Seachem AS and Jo Tankers B.V., instituted an internal investigation and received a determination that the company’s conduct was anticompetitive.

Being the first member of the cartel to report the anticompetitive arrangement, Stolt-Nielsen executed a non-prosecution agreement with the government on January 15, 2003. Under the agreement, Stolt-Nielsen agreed to cooperate with the government. In exchange, the government promised “not to bring any criminal prosecution against [Stolt-Nielsen] for any act or offense it may have committed prior to [January 15, 2003] in connection with the anticompetitive activity being reported.” Subsequently, Stolt-Nielson provided the government with highly incriminating evidence concerning the customer allocation conspiracy. This evidence eventually led to successful prosecutions of Odfjell, Jo Tankers, and several of their respective executives; the government collected a total of $62 million in fines and sent three high-ranking officials to jail.

Approximately three months after the agreement was reached, the government suspended it claiming that while Stolt-Nielsen learned of its illegal conduct in early 2002, it continued that conduct into November 2002, thereby materially breaching the amnesty agreement which specified that the company “took prompt and effective action to terminate its part in the anticompetitive activity being reported upon discovery of the activity.” Thereafter, the government arrested Stolt-Nielsen’s managing director.

Stolt-Nielsen immediately sought to enjoin the Antitrust Division from seeking an indictment against the company and its officers, relying on the principles of contract law that it had fulfilled its obligation and thus deserved the benefit of the bargain. It also argued that the agreement was silent as to when the illegal conduct had ceased. After a hearing, Judge Timothy J. Savage of the District Court for the Eastern District of Pennsylvania granted the injunction, concluding that Stolt-Nielsen had not breached the amnesty agreement. The United States Court of Appeals for the Third Circuit reversed the district court on separation of powers grounds, stating that unless and until the Division obtained an indictment, the district court could not determine the issue. After the government indicted Stolt-Nielsen, its affiliate and several of its officers, the defendants moved to dismiss the indictment.

A Second District Court Judge Dismisses the Indictment

After holding an evidentiary hearing, Judge Bruce W. Kauffman, the second district court judge to hear the issue, dismissed the indictment. He found that under the terms of the amnesty agreement, Stolt-Nielsen did not promise to terminate all anticompetitive conduct by March 2002, as the government contended, but instead committed to institute “a prompt and diligent process” to stop the activity. The court further found that Stolt-Nielsen had complied with this obligation because by March 2002 it had: 

  • Instituted a comprehensive Antitrust Compliance Policy documented in a revised Antitrust Compliance Handbook; 
  • Promptly distributed the Handbook to employees and competitors; 
  • Held a series of mandatory seminars at Stolt-Nielsen’s offices around the world, attended by top management, to inform its executives and employees of the necessity of antitrust compliance; 
  • Required all relevant employees to sign certifications in which they represented that they would comply strictly with all terms of the Antitrust Compliance Policy or risk demotion or termination; and 
  • Informed its competitors of the revised Policy and the company’s intention to comply with it.

The Court expressly rejected the government’s claim that the illegal activity had continued, criticizing the government’s co-conspirator witnesses as inherently untrustworthy as they were the ones whom Stolt-Nielsen had originally reported to the Antitrust Division. Moreover, the Court concluded that after self-reporting it made no common sense for the company and its officers to continue the conspiracy. In addition, the Court found that contrary to the Division’s position, Stolt-Nielsen had begun competing vigorously in March 2002, including on three contracts the Division had identified as representing further collusion. The Court also criticized the Division’s drafting of the agreement, which did not define “prompt and effective action,” so applying fundamental principles of contract law, any ambiguity was resolved in favor of the defendants. For these reasons, the court ultimately concluded that Stolt-Nielsen did not breach the amnesty agreement and that the government was bound by the agreement’s terms.


The Stolt-Nielsen case is significant for several reasons. First and foremost, the case demonstrates that participation in the Corporate Leniency Program remains a legitimate option for potential antitrust defendants. Had the case been decided differently, many would have questioned the value of the amnesty program. Now firms considering entering into amnesty agreements have the comfort that disputes will not be decided unilaterally by the Division, but instead will be decided by the courts. Second, the decision reaffirms the general principle that criminal non-prosecution agreements are evaluated under contract law and that as the government drafts the agreements, any ambiguities are construed against the government. Practitioners should expect, therefore, that the government will be much more exacting in drafting these agreements in the future. Finally, procedural changes to the program have already begun; notably, the Division is less willing to rely on counsel’s proffer for a grant of conditional amnesty, requiring instead verification by actual evidence. Consequently, the process of obtaining amnesty will take more time than it used to, but the chance of being rejected once in the program is more remote.