dissenting in part.
I am pleased to join Judge Stapleton’s opinion for the Court in most respects. Unlike my colleagues, however, I would hold that Novartis had no duty to test AG600 for tank-mixing compatibility with the fungicides used by Plaintiffs. I therefore would affirm the District Court’s summary judgment on Plaintiffs’ design defect claim under the New Jersey Products Liability Act. And because it follows a fortiori that Novartis had no duty to warn Plaintiffs about the results of a test it had no legal duty to conduct, I would affirm the District Court’s summary judgment on Plaintiffs’ implied failure-to-warn claim as well. Accordingly, I respectfully dissent from Parts IV and V of the Court’s opinion.
I.
The extensive factual background and tortuous procedural history of this case are aptly recounted in Judge Stapleton’s thorough opinion, so I shall briefly mention only a few points relevant to my dissent.
Plaintiffs’ fourth amended complaint includes two claims brought pursuant to the New Jersey Products Liability Act (NJPLA), N.J. Stat. Ann. § 2A:58C-1 et seq. First, Plaintiffs allege that AG600 was defectively designed because it contained an undisclosed ionic surfactant that harmed Plaintiffs’ crops when “tank mixed” with certain fungicides regularly used by Plaintiffs, such as Captan and Captec. Although Novartis did not recommend—and in fact warned against—combining AG600 with fungicides during the application process, Plaintiffs contend that Novartis had a duty to test AG600 for adverse interactions with other agricultural chemicals before distributing it. Second, Plaintiffs aver that AG600 was defective because Novartis failed to warn of the dangers inherent in tank mixing the pesticide with Plaintiffs’ fungicides.
In Parts IV and V of its opinion, the Court holds that the District Court erred in granting Novartis summary judgment on both of Plaintiffs’ NJPLA claims. Because I believe Plaintiffs’ practice of combining AG600 with various fungicides during the tank mixing process was not objectively foreseeable to Novartis, I disagree.
A.
I begin with Plaintiffs’■ claim for defective design. Under the NJPLA, a plaintiff asserting a claim for defective design must prove by a preponderance of the evidence that “the product causing the harm was not reasonably fit, suitable or safe for its intended purpose because it ... was designed in a defective manner.” N.J. Stat. Ann. § 2A:58C-2. When a product is used for something other than its specifically intended purpose, a plaintiff nonetheless may prevail on a design defect claim under the NJPLA by demonstrating that such use was “reasonably foreseeable” to the manufacturer. Jurado v. Western Gear Works, 131 N.J. 375, 619 A.2d 1312, 1317 (1993) (citations and quotation marks omitted). Because it is undisputed that Novartis designed AG600 as a stand-alone pesticide and did not recommend applying it in conjunction with fungicides, Plaintiffs bear the burden of demonstrating that their tank mixing practice was a reasonably *230foreseeable misuse of the pesticide. Id. at 1317-18.1
To determine whether the misuse of a product was reasonably foreseeable under the NJPLA, we apply an objective test. Id. at 1317. We do not ask whether a manufacturer was' aware of previous instances in which its product had been similarly misused because such evidence “tends to show only subjective foreseeability,” which is irrelevant to the objective foreseeability analysis. Port Auth. of New York and New Jersey v. Arcadian Corp., 189 F.3d 305, 315 (3d Cir.1999) (quoting Oquendo v. Bettcher Indus., Inc., 939 F.Supp. 357, 363 (D.N.J.1996)); see also Brown v. U.S. Stove Co., 98 N.J. 155, 484 A.2d 1234, 1241 (1984). Instead, whether the misuse of a product is objectively foreseeable—and thereby imposes a duty on the manufacturer to take steps to ensure that the product is safe for that use—is ultimately “a question of fairness and public policy.” Arcadian, 189 F.3d at 315. For that reason, we have emphasized that “[fjairness, not foreseeability alone, is the test” for reasonable foreseeability under the NJPLA. Id. at 316 (quoting Kuzmicz v. Ivy Hill Park Apartments, Inc., 147 N.J. 510, 688 A.2d 1018, 1020 (1997)).
It is in this assessment of the objective foreseeability of Plaintiffs’ tank mixing of AG600 with fungicides where I part ways from my colleagues. The Majority marshals a considerable amount of evidence to support its conclusion that tank mixing was a reasonably foreseeable misuse of AG600. According to the Majority, individual plaintiff farmers, pesticide dealers, and Rutgers University crop-treatment scientists all testified that tank mixing pesticides with fungicides was a common and well-known industry practice among blueberry farmers. Furthermore, Novartis sales representatives themselves acknowledged a general awareness that farmers often engage in tank mixing.
Although such evidence suggests a subjective awareness on the part of Novartis and others that Plaintiffs would tank mix AG600 with different fungicides, it sheds no light on the question of whether this subjectively foreseeable misuse is objectively reasonable. For a duty to attach under the NJPLA, the misuse of a product must be objectively foreseeable to the manufacturer—a determination which, as explained above, turns on questions of fairness and public policy, not Novartis’s subjective awareness of past tank mixing by Plaintiffs. See Jurado, 619 A.2d at 1317; Arcadian, 189 F.3d at 315. Because the deposition testimony cited by the Majority does not indicate whether it would be either fair or sound public policy to charge Novartis with responsibility for ensuring that Plaintiffs’ practice of tank mixing AG600 with various fungicides was safe, that evidence would seem irrelevant to the question of whether Plaintiffs’ misuse of AG600 was objectively foreseeable under the NJPLA.
Unlike the Majority, I am not convinced that Plaintiffs have carried their burden of demonstrating that their misuse of AG600 was reasonably foreseeable to Novartis. As the Majority’s foreseeability analysis demonstrates, Plaintiffs can point to no relevant evidence in the record that suggests tank mixing was an objectively foreseeable misuse of AG600. Instead, Plaintiffs—like the Majority-—-rely almost entirely on evidence of subjective foreseeability. Because the burden of demonstrating reasonable foreseeability rests on *231Plaintiffs, see Jurado, 619 A.2d at 1317, I would affirm the District Court’s conclusion that Novartis had no duty under the NJPLA to test AG600 for compatibility with various fungicides.
In contrast to Plaintiffs’ failure to introduce relevant evidence on the issue of reasonable foreseeability, Novartis has cited persuasive evidence which strongly suggests that it would be neither fair nor prudent public policy to impose a duty on Novartis to test AG600 for compatibility with fungicides such as Captan and Captec. When viewed as a whole, this evidence compels the conclusion that Plaintiffs’ practice of tank mixing was not an objectively foreseeable misuse of AG600.
Evidence suggests that the burden such testing would impose on Novartis would be substantial to say the least. The AG600 label indicates that the pesticide is approved for use not just on the blueberries that Plaintiffs raise but on 62 different types of plants. See App. at 548-574. Novartis has represented—and Plaintiffs have not disagreed—that there are approximately 98 different registered fungicides and 141 different registered insecticides that growers of those 62 types of plants could elect to tank mix with AG600. Appellee’s Br. at 13-14, 27-28. This yields over 850,000 tank-mix/plant combinations that Novartis would be required to test for compatibility before marketing AG600.2 Evidence also indicates that such extensive testing would be time consuming: a single test conducted by Rutgers University scientists involving AG600, Captan, Captec, and blueberry plants took two full years to complete. See App. at 614-620.3
The Majority points to the deposition testimony of Dr. James Witt, who testified that Novartis need not test AG600 for every possible tank-mix combination. Instead, Dr. Witt opined that Novartis could have looked to “what some of the common and usual practices are, and the areas where [AG 600] is going to be sold and used” and limited its testing accordingly. App. at 764-765. According to the Majority, Dr. Witt’s testimony suggests the testing burden is not nearly as onerous as Novartis claims.
With additional information, Dr. Witt’s testimony could be a helpful means of evaluating the testing burden that the Majority’s holding imposes on Novartis. Despite their burden of demonstrating the objective foreseeability of tank mixing, however, Plaintiffs have cited no additional evidence indicating where AG600 is commonly used and sold and no evidence suggesting what the “common and usual practices” of farmers who use AG600 actually are. Without such information, Dr. Witt’s testimony provides us with no basis for concluding that the Majority’s holding will require Novartis to test anything less than the approximately 850,000 potential tank-mix combinations discussed above. The notion that *232far fewer tests would be required is based, I believe, on the fallacy that the manufacturer knows in advance which of the many potential AG600/insecticide/fungicide/crop combinations the end user will choose.
In my view, the considerable burden that the Majority’s holding will impose on manufacturers is unsound public policy. Requiring Novartis to test AG600—and, apparently, all other pesticides—in innumerable tank mixing combinations would stifle the development of agricultural pesticides and increase substantially their cost of production. This would, in turn, drive up the cost of food, since pesticide manufacturers and farmers would inevitably pass on at least a portion of their escalating costs to consumers. As a matter of policy, then, it seems both logical and prudent to lay the responsibility for ensuring tank-mixing compatibility on end users such as Plaintiffs, who have actual knowledge of the specific types of fungicides that they wish to combine with AG600 and the various crops they wish to treat with the mixture.
Nor would requiring Novartis to test AG600 for tank mix compatibility be particularly fair, either. Because the Environmental Protection Agency has not approved AG600 for mixing with fungicides such as Captan and Captec, any such mixture would be considered an “off-label” use of the pesticide. Federal law prohibits Novartis from recommending or marketing AG600 for off-label uses. See 7 U.S.C. § 136j(a)(1)(B). Furthermore, Novartis explicitly cautions farmers against mixing AG600 with other chemical substances. See App. at 545. Accordingly, the Majority’s holding will require Novartis to undertake extensive and expensive testing to ensure that AG600 is fit to be used in a manner that it warns against and is explicitly prohibited from advocating.
In sum, the record evidence suggests that requiring Novartis to test AG600 for tank mix compatibility with fungicides such as Captan and Captec is both unfair and unsound as a matter of public policy. Accordingly, I would conclude that Plaintiffs misuse of AG600 in the tank mixing process was not objectively foreseeable to Novartis, despite Plaintiffs’ evidence suggesting that the company was subjectively aware of the practice. Because Novartis had no duty to test AG600 for compatibility with the fungicides employed by Plaintiffs, I would affirm the District Court’s summary judgment on Plaintiffs’ defective design claim under the NJPLA.
B.
The District Court also granted summary judgment on Plaintiffs’ failure-to-warn claim brought pursuant to the NJPLA after concluding that the claim was preempted. In Part IV of its thorough opinion, the Majority persuasively explains why this holding was error, and I agree fully with its preemption analysis.
Notwithstanding the District Court’s flawed preemption reasoning, I do not believe it necessary to reverse the District Court’s dismissal of Plaintiffs’ failure-to-warn claim. For the reasons explained herein, I would hold that Novartis had no duty to test AG600 for tank mix compatibility with the fungicides used by Plaintiffs. It follows a fortiori that Novartis had no duty to warn Plaintiffs. Accordingly, I would affirm the District Court’s summary judgment on Plaintiffs’ implied failuré-to-warn claim as well.
. Under the NJPLA,' use of a product for anything other than its intended purpose is termed “misuse.” See Jurado v. Western Gear Works, 131 N.J. 375, 619 A.2d 1312, 1317 (1993).
. Novartis cites a higher figure, claiming that the Majority's holding will require it to test "at least 2,963,220 different three-product use combinations'' before distributing AG600. See Appellee's Br. at 27. Novartis’s calculation seems to assume erroneously that farmers will combine AG600 with multiple fungicides and multiple insecticides at the same time. The record, however, indicates only that farmers will combine AG600 with, at most, one fungicide and one insecticide during any given application. Accordingly, I use the smaller figure of 850,000 as a more accurate reflection of the testing burden facing Novartis.
. The Majority notes that Plaintiffs introduced evidence suggesting that Novartis could detect potential compatibility problems with a simple "jar test” of AG600 and a given fungicide. Accepting this as true, I believe it would be an unreasonable burden to require a manufacturer such as Novartis to perform over 850,000 separate jar tests.