GILMAN, J., delivered the opinion of the court, in which MOORE, J., joined. NATHANIEL R. JONES, J. (pp. 1052-53), delivered a separate concurring opinion.
OPINION
GILMAN, Circuit Judge.The Kalamazoo River Study Group (KRSG), an unincorporated association of paper manufacturers, brought suit in federal district court pursuant to the Comprehensive Environmental Response, Compensation and Liability Act of 1980 (CERCLA), 42 U.S.C. §§ 9601-75, seeking contribution from Rockwell International Corporation for the latter’s role in contaminating the Kalamazoo River with polychlo-rinated biphenyls (PCBs). In a bifurcated bench trial, the district court first determined that Rockwell’s release of PCBs into the Kalamazoo, River was significant enough for it to face liability under CERC-LA. But the district court ultimately declined to allocate any. response costs to Rockwell, finding that its release of PCBs was minuscule (less than one-hundreth of 1%) in comparison with that of the companies comprising the KRSG. For the reasons set forth below, we AFFIRM the judgment of the district court;
I. BACKGROUND
This case arises from' the presence of PCBs in a portion of the Kalamazoo River located in the state of Michigan. The substance is a synthetic liquid with many industrial uses. It is also a hazardous material that poses significant health and environmental risks. Because of these risks, the manufacture of PCBs ceased in the 1970s. At approximately the same time, the state agency now known as the Michigan Department of Environmental Quality (MDEQ) began studying" the level of PCBs in the Kalamazoo River. The MDEQ completed its initial investigation in 1990, concluding that a 35 mile stretch of the River was contaminated with PCBs. This stretch begins at the confluence of the Kalamazoo River with Portage Creek, and continues downstream to the Allegan City Dam.
Based upon the findings of the MDEQ, the United States Environmental Protection Agency (EPA) placed this portion of the River, along with a three-mile portion of Portage Creek, on the National Priorities List as a Superfund Site pursuant to § 105 of CERCLA (42 U.S.C. § 9605) (collectively, the Site). The EPA subsequently authorized the MDEQ to conduct an EndangermenVRisk Assessment (E/RA) of the Site. Following the E/RA, the MDEQ identified three paper mills as be*1046ing potentially responsible for the PCB contamination: Georgia Pacific Corporation, Millennium Holdings, Incorporated, and Plainwell, Incorporated. These companies then entered into an Administrative Order by Consent (AOC) that required them to fund a Remedial Investigation and Feasibility Study (RI/FS) at the Site and its surrounding area. Fort James Operating Company later agreed to share the costs of the RI/FS, joining with the other companies to form the KRSG.
Pursuant to the AOC, the RI/FS encompassed a 95 mile stretch of the Kalamazoo River running both upstream and downstream from the Site. This expanded area included the portion of the River that is adjacent to the former site of Rockwell’s manufacturing facility in Allegan, Michigan. From approximately 1910 to 1989, Rockwell built universal joints for the automotive industry at its Allegan facility.
In 1995, the KRSG brought suit against Rockwell and several other companies in the United States District Court for the Western District of Michigan. The KRSG alleged that these companies were partly responsible for contaminating the Site with PCBs. It therefore sought contribution from them for the costs associated with both the RI/FS and the future clean-up of the Site. Although the KRSG asserted various bases for its right to contribution, the district court and the parties focused exclusively on the KRSG’s contribution claim pursuant to § 113(f) of CERCLA (42 U.S.C. § 9613(f)). The KRSG’s contribution claims against the other companies subsequently settled or were otherwise resolved, leaving only its claim for contribution against Rockwell for resolution by the district court.
A procedural ruling by the district court bifurcated the trial of the KRSG’s contribution claim against Rockwell into two stages, with the first limited to liability and the second focused on the allocation of response costs. Both stages were tried to the bench. At the liability stage, the district court employed a “threshold of significance” standard of liability, a standard later rejected by this court. As articulated by the district court, this standard imposed CERCLA liability where a defendant’s release of hazardous material is of sufficient significance to justify response costs. Kalamazoo River Study Group v. Menasha Corp., 228 F.3d 648, 654 (6th Cir.2000) (describing the threshold of significance standard). The district court determined that the KRSG and Rockwell had both released a sufficient amount of PCBs to face liability under the threshold of significance standard. It observed, however, that Rockwell’s release of PCBs appeared to be minimal in comparison to the release of PCBs by the members of the KRSG. Although this court later rejected the threshold of significance standard because it improperly requires the plaintiff to show that a defendant’s release of hazardous materials caused response costs, see id. at 655, the adoption of a lower liability standard did not inure to the benefit of Rockwell. The district court’s finding that Rockwell had released a sufficient amount of PCBs to be held potentially liable even under the more onerous threshold of significance standard would obviously not change when subjected to the lower standard.
Following the liability stage, the district court considered the proper allocation of response costs between the KRSG and Rockwell. The district court identified three factors as generally relevant to the allocation of response costs: (1) the relative quantities of PCBs released by the parties, (2) the relative toxicity of those PCBs, and (3) the cooperation of the parties with the regulatory authorities. After the court found that the latter two factors did not favor any particular allocation of response costs, it focused on the relative quantity of PCBs released by Rockwell *1047versus the amount released by the KRSG. The district court determined that Rockwell had likely released no more than 20 pounds of PCBs from its Allegan facility. In contrast, the court found that the members of the KRSG had released “hundreds of thousands of pounds” of PCBs into the River. Based upon these findings, the district court did not allocate any response costs to Rockwell. The KRSG now appeals the district court’s decision.
II. ANALYSIS
A. Standard of review
A district court’s allocation of response costs in a CERCLA contribution action will not be set aside unless we determine that the court abused its discretion. United States v. R.W. Meyer, Inc., 932 F.2d 568, 573 (6th Cir.1991). An abuse of discretion is found where we are left with the “definite and firm conviction that the trial court committed a clear error of judgment.” Logan v. Dayton Hudson Corp., 865 F.2d 789, 790 (6th Cir.1989).
The factual findings underlying the district court’s allocation of response costs may be set aside only if clearly erroneous. Schroyer v. Frankel, 197 F.3d 1170, 1173 (6th Cir.1999). A factual finding is clearly erroneous where, although there is evidence to support that finding, “the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 92 L.Ed. 746 (1948).
B. The district court did not abuse its discretion in declining to allocate response costs to Rockwell
1. A holding of potential liability does not preclude a zero allocation of response costs
The KRSG argues that the district court’s refusal to allocate response costs to Rockwell is inconsistent with its earlier conclusion that Rockwell faced liability under CERLCA for releasing PCBs into the Kalamazoo River. Specifically, the KRSG maintains that the district court cannot logically decline to allocate response costs to Rockwell after determining that it faced liability under the now-discredited threshold of significance standard. We disagree.
At the allocation stage of the trial, the district court focused on the relative quantities of PCBs released into the Kalamazoo River by the parties. But in determining that Rockwell faced liability under CERC-LA, the district court did not make specific findings with regard to the amount of PCBs released by Rockwell versus the amount released by the KRSG. The district court explicitly stated that, at the liability stage, it was “not called upon to quantify Rockwell’s release of PCBs to the River.” It instead focused on whether Rockwell’s release of PCBs was “more than incidental or sporadic.” The district court ultimately concluded that Rockwell faced liability under CERCLA after finding that Rockwell released PCBs in “measurable or detectable quantities.” This finding did not obligate the district court to allocate response costs to Rockwell irrespective of the court’s specific analysis of the relative amount of PCBs released by Rockwell versus the KRSG.
The United States Court of Appeals for the Seventh Circuit faced this very issue in PMC, Inc. v. Sherwin-Williams Co., 151 F.3d 610 (7th Cir.1998). In PMC, the Seventh Circuit held that the district court did not abuse its discretion in declining to allocate response costs to a polluter who admitted to dumping toxic waste. Id. at 616. The court explained that the polluter’s “spills may have been too inconsequential to affect the cost of cleaning up significantly, and in that event a zero allocation to [the polluter] would be appropri*1048ate.” Id. As in the case before us, the other polluter in PMC was found responsible for substantially all of the total contamination of the site. Id.
Seeking to distinguish the PMC decision, the KRSG points out that the court in PMC used a standard that imposed liability no matter how small the release of hazardous material. A liability determination under this standard does not necessarily require an allocation of response costs, according to the KRSG, because liability may be imposed absent a finding of any significant release. Under the threshold of significance standard of liability mistakenly employed by the district court at the liability phase in the present case, however, the KRSG argues that a determination of liability necessarily means that the court found that the defendant had released a significant amount of hazardous material.
The KRSG misses the mark, however, because the court in PMC was not concerned with whether the polluter had released a significant amount of hazardous material. Instead, the court looked to whether the polluter’s release of hazardous material was too inconsequential in comparison to that of the other polluter to significantly affect clean-up costs. Id. In other words, where the other responsible parties release vast quantities of hazardous material, a defendant’s release of what, standing alone, would be a significant amount of such material might have no impact on the total cost of cleaning up a contaminated site.
This is not to say that a defendant can always avoid paying response costs where its release does not significantly affect clean-up costs. If, for example, all of the responsible parties have each released only a relatively small amount of hazardous material, then each individual release in isolation would have little impact on the total cost of cleaning up a contaminated site. Nevertheless, a court faced with these circumstances could reasonably allocate a portion of the response costs to each party. But this is not the situation in the present case. The district court concluded that the companies comprising the KRSG each released exponentially more PCBs into the Kalamazoo River than Rockwell, so that Rockwell’s release will have essentially no effect on the as yet-undetermined clean-up costs.
Even assuming that the district court’s liability determination did not require an allocation of future clean-up costs to Rockwell, the KRSG argues that this determination should have at least led the district court to require Rockwell to pay for the some of the costs associated with the RI/ FS. These costs, according to the KRSG, should be allocated to Rockwell even if it released a relatively small amount of PCBs into the River. Specifically, the KRSG argues that CERCLA authorizes the allocation of investigation costs to any party that created a reasonable risk of contaminating a site.
In support of its argument, the KRSG cites Johnson v. James Langley Operating Co., 226 F.3d 957, 964 (8th Cir.2000) (stating that a plaintiff may recover the costs associated with environmental testing or sampling “only if the party seeking to recover costs has an objectively reasonable belief that the defendant’s release or threatened release of hazardous substances would contaminate his or her property”), and Lansford-Coaldale Joint Water Auth. v. Tonolli Corp., 4 F.3d 1209, 1219 (3d Cir.1993) (stating that CERCLA liability for environmental investigation costs requires, among other things, that “there was a reasonable risk (although one that may not materialize) that the defendant’s release or threatened release of hazardous substances would contaminate the plaintiffs property”).
*1049Both Johnson and Lansford-Coaldale address the showing required to establish a party’s liability for investigation costs in an action brought by a landowner pursuant to § 107(a) of CERCLA (42 U.S.C. § 9607(a)). A liability determination, however, is just the first element of a contribution claim under § 113(f). “Recovery of response costs by a private party under CERCLA is a two-step process. Initially, a plaintiff must prove that a defendant is liable under CERCLA. Once that is accomplished, the defendant’s share of liability is apportioned in an equitable manner.” Kalamazoo River Study Group v. Menasha Corp., 228 F.3d 648, 656-57 (6th Cir.2000) (internal quotation marks and citation omitted). Neither Johnson nor Lansford-Coaldale hold that a defendant who is subject to liability for investigation costs must necessarily be allocated a share of those costs in a contribution action. Accordingly, these cases provide no guidance as to the proper allocation of such costs in the present case.
The district court has broad discretion to allocate the costs associated with the RI/FS. Franklin County Convention Facilities Auth. v. Am. Premier Underwriters, Inc., 240 F.3d 534, 549 (6th Cir.2001) (“The apportionment of CERCLA liability under § 113(f) among various responsible parties is an equitable undertaking within the broad discretion of the district court.”). In allocating these costs, the district court is authorized to consider any “equitable factors” that it considered “appropriate.” 42 U.S.C. § 9613(f). The district court’s decision not to allocate any costs for the RI/FS to Rockwell was based upon its finding that the KRSG was responsible for more than 99.9% of the PCBs in- the River. Although the KRSG challenges this finding, a challenge that we address in Part II.B.2. below, it fails to show that the district court abused its discretion in looking to the relative quantities of PCBs released by the parties in allocating costs for the RI/FS.
The KRSG further argues that the district court’s failure to allocate response costs to Rockwell after finding that it had released PCBs into the Kalamazoo River defeats the central purpose of CERCLA; namely, the prompt clean-up of hazardous waste. According to the KRSG, the district courtls allocation of response costs in this case encourages parties to litigate “in the hope of obtaining a zero share, rather than voluntarily joining in the investigation or settling.” But the allocation of response costs is highly fact-intensive, so that an allocation ,of zero response costs in a particular case provides little incentive for defendants in other contribution actions to reject reasonable settlement offers or risk the uncertainties inherent in litigation,
For all of these reasons, we conclude that the district court’s liability determination did not obligate it to allocate response costs to Rockwell.
2. The district coiirt did not err in finding that Rockwell had released an inconsequential amount of PCBs in comparison to the amount of PCBs released by the members of the KRSG
The relative quantities of PCBs released by the parties was the decisive factor in the district court’s allocation of response costs. It found that Rockwell had likely released less than 20 pounds of PCBs into the Kalamazoo River. In contrast, the district court determined that the KRSG members had released several hundred thousand pounds of PCBs into the River. The KRSG concedes that its own members released massive amounts of PCBs, but maintains that the district court erred in concluding that Rockwell had re*1050leased such a small amount of the hazardous substance.
In assessing Rockwell’s release of PCBs, the district court gave credence to the testimony of Robert Barrick, an expert in environmental chemistry. Barrick testified that he formed an opinion as to the amount of PCBs that Rockwell released into the Kalamazoo River by analyzing estimates of both the amount of oil that Rockwell had discharged and the concentration of PCBs in those oils. With regard to the amount of discharged oils, Barrick used the estimate offered by Dr. Kenneth Crumrine, the KRSG’s expert. Barrick then estimated the concentration of PCBs in those oils by examining the oils remaining in the groundwater at the site of Rockwell’s Allegan facility. He determined that these oils contained no more than 0.000035% PCB. Based upon his analysis of these two estimates, Barrick concluded that Rockwell had likely released less than 20 pounds of PCBs into the River.
The KRSG challenges Barrick’s opinion on several grounds. First, the KRSG contends that Barrick could not accurately estimate the amount of PCBs released by Rockwell without having the expertise to predict how the oils discharged from Rockwell’s facility would have reacted once in the River. We find no merit in this argument, however, because the KRSG fails to explain why Barrick needed to possess such expertise in order to form a reliable opinion as to Rockwell’s release of PCBs. The mathematical methodology employed by Barrick, as well as by Dr. Crumrine, requires an assessment of only the amount of discharged oil and the concentration of PCBs in that oil. This methodology requires no analysis of how PCBs travel or change in a river environment.
The KRSG further claims that Barrick had no basis for concluding that the concentration of PCBs in the oil discharged by Rockwell was the same as the concentration in the oil currently found in the groundwater at the site of the Allegan facility. But KRSG did not challenge Bar-rick’s testimony on this ground at trial. Furthermore, Barrick testified that his analysis of the oil in the groundwater revealed very little evidence of any weathering or degradation, thus demonstrating that the PCB concentration of the oil in the groundwater was representative of the concentration in the oils discharged by Rockwell. The KRSG offered no evidence to refute this testimony.
Next, the KRSG maintains that the district court erred in declining to accept Dr. Crumrine’s estimate of the PCB concentration in the oils discharged by Rockwell. Dr. Crumrine estimated that these oils contained either 5% or 50% PCB, depending on the particular type of oil. The district court reasonably rejected this estimate based upon Barrick’s testimony that it is “physically impossible” for oils containing 5% or 50% PCB to be reduced to a PCB concentration of only 0.000035%, the concentration of the oil found in the groundwater at the site of the Allegan facility. In addition, the district court pointed out that Dr. Crumrine’s estimate failed to take into account that, beginning in the early 1960s, Rockwell increasingly used water-soluble oils that might not have contained PCBs at all.
Finally, the KRSG argues that Barrick’s opinion is rebutted by other evidence showing that Rockwell in fact released a large amount of PCBs -into the River. The KRSG specifically relies upon a few sediment samples gathered from the River that contained elevated levels of Aroclor 1254, the type of PCB that Rockwell used at its Allegan facility. One such sample, identified as “BR 27,” was recovered 1.7 miles from the Allegan facility and contained a very high level of Aroclor 1254. The KRSG contends that BR-27, as well *1051as six other sediment samples with high Aroclor 1254 levels, prove that Rockwell released a large amount of PCBs into the River.
In our view, the district court properly determined that these samples were of limited probative value. As the district court pointed out, Barrick gathered approximately 300 sediment samples from “areas of the river in which oils would be expected to accumulate downstream of Rockwell.” Only seven of the samples contained high levels of Aroclor 1254. These samples, comprising less than 3% of the total number of samples removed from the River, were apparent anomalies that neither party could explain. Their presence does not discredit Barrick’s opinion regarding the amount of PCBs released by Rockwell.
Based on all of the above, we conclude that the district court’s factual determination that Rockwell likely released less than 20 pounds of PCBs into the Kalamazoo River is not clearly erroneous. We therefore need not address the KRSG’s challenge to the other evidence that the district court cited as corroborating Barrick’s opinion.
3. The district court did not err in determining that the factors concerning the relative toxicity of the PCBs released by the parties and the cooperation of the parties with the regulatory authorities did not favor any particular allocation of response costs
The district court recognized that, in general, the relative toxicity of the PCBs released by the parties and the parties’ cooperation with the regulatory authorities are both relevant factors in allocating response costs. It determined, however, that neither factor offered any guidance as to the proper allocation of response costs in the present case. The KRSG challenges this determination, arguing that both factors favor allocating response costs to Rockwell.
First, the KRSG contends that the district court erroneously found that Rockwell and the KRSG members had released PCBs of approximately the same toxicity. The KRSG maintains that Aroclor 1254, the type of PCB that Rockwell released into the Kalamazoo River, is more toxic than Aroclor 1242, the type of PCB that its members released into the River. According to the KRSG, Aroclor 1254 bioaccumu-lates in fish at a much higher rate than Aroclor 1242, a fact that the KRSG insists is significant because the concerns about PCB levels in fish are allegedly “driving the response in this case.” The KRSG further argues that, in terms of carcinogenic risk, the EPA considers Aroclor 1254 more toxic than Aroclor 1242.
But the district court had a reasonable basis for treating Aroclor 1254 and 1242 as equally toxic. In particular, the MDEQ issues fish advisories and other regulatory criteria without distinguishing between the different types of PCBs. The court noted that the MDEQ treats all PCBs the same because every type of PCB contains toxins. Although the evidence presented by the KRSG adequately supports a finding that Aroclor 1254 is more toxic than Aroclor 1242, we are not left with a “definite and firm conviction” that the district court erred in following the approach of the MDEQ. United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 92 L.Ed. 746 (1948). Moreover, in light of the district court’s finding with regard to the huge disparity in the relative quantities of PCBs released by the parties, a determination that Aroclor 1254 is somewhat more toxic than Aroclor 1242 would not likely have altered the court’s allocation of response costs.
*1052The KRSG next argues that the district court’s consideration of the cooperation factor was “deficient.” Specifically, the KRSG points to evidence showing that Rockwell did not fully cooperate with the regulatory authorities. Rockwell, according to the KRSG, failed to provide important data to these authorities and “contrived stories” in an attempt to “explain away” its responsibility for releasing PCBs into the Kalamazoo River.
The district court, however, in fact recognized that Rockwell had not fully cooperated with the regulatory' authorities. But the court nevertheless determined that the cooperation factor did not weigh in favor of the KRSG because it found “a lack of full cooperation by both parties.” (Emphasis added.) The KRSG offers no rebuttal to the district court’s determination that, like Rockwell, it too did not fully cooperate with the regulatory authorities.
Accordingly, we conclude that the district court reasonably determined that both the toxicity and cooperation factors were not determinative in the allocation of response costs in the present case.
III. CONCLUSION
For all of the reasons set forth above, we AFFIRM the judgment of the district court.