concurring.
In my opinion defendant, United States Stove Company, owed no duty to plaintiff and, even if it did, the asserted breach of that duty was not proven to be a cause-in-fact of the breach. The majority, however, asserts that defendant owed a duty to plaintiff. It then proceeds to intertwine concepts of duty with causation-in-fact and concludes that there was no jury issue with respect to the issue of proximate cause. The result is correct, but I am compelled to write because of our differences on the notions of duty and cause-in-fact.
These differences can perhaps be best understood by analyzing the functions that a trial court performs in a negligence suit. First, the trial court must decide whether the injured interest is one that falls within the protection of the rule invoked by the plaintiff. In other words, was the defendant’s conduct a hazard against which the rule afforded protection? It is in this sense that a judicial determination must be made whether a duty exists.
Examples of this type of inquiry may be found in Goldberg v. Housing Auth. of Newark, 38 N.J. 578 (1962), and Caputzal v. The Lindsay Co., 48 N.J. 69 (1966). In Goldberg, the plaintiff was beaten by two men in a housing project owned by the *176defendant. There was evidence that the defendant knew that substantial criminal activity had been occurring on the premises. Chief Justice Weintraub’s analysis pointed out that the question of whether a duty existed could not be resolved by recourse to “foreseeability.” Rather, the question was whether a duty existed to take measures to guard against criminal acts. He then continued:
Whether a duty exists is ultimately a question of fairness. The inquiry involves a weighing of the relationship of the parties, the nature of the risk, and public interest in the proposed solution. [38 N.J. at 583.]
Caputzal was a strict liability action. Plaintiff sued the manufacturer of a water softener and the company that sold it to him, alleging that it was defective in either its manufacture or installation. The defect caused the water to become discolored and, when the plaintiff discovered that he had drunk the discolored water, he suffered a heart attack. This Court upheld a summary judgment for the defendants, though all the facts stated in the complaint, including causation-in-fact, were assumed to be true. Justice Hall, writing for a unanimous Court, stated:
Foreseeability is not solely a mere matter of logic, since anything is foreseeable, but frequently involves questions of policy as well. When it does, the matter is one for determination by the court and not by the factfinder.
In the instant case it is much too fanciful to say, from the point of view of fairness, that a reasonable manufacturer, seller or installer of water softeners should be held to recognize that he would create an unreasonable risk of any substantial physical injury whatever to anyone by the defect here involved, let alone a heart attack. [48 N.J. at 75-76 (emphasis added).]
Justice Hall further observed:
Recovery for the type of result here involved is denied in many cases also on the ground of lack of proximate cause, which we should briefly mention. Utilization of that term to draw judicial lines beyond which liability will not be extended is fundamentally as an instrument of fairness and policy, although the conclusion is frequently expressed in the confusing language of causation, “foreseeability” and “natural and probable consequences.” Many years ago a case in this State hit it on the head when it was said that the determination of proximate cause by a court is to be based “ 'upon mixed considerations of logic, common sense, justice, policy and precedent.’ ” Powers v. Standard Oil Co., *17798 N.J.L. 730, 734 (Sup.Ct.1923), affirmed o.b. 98 N.J.L. 893 (E. & A.1923). Dean Prosser puts it well in this fashion:
“As a practical matter, legal responsibility must be limited to those causes which are so closely connected with the result and of such significance that the law is justified in imposing liability. Some boundary must be set to liability for the consequences of any act, upon the basis of some social idea of justice or policy.
This limitation is sometimes, although rather infrequently, one of the fact of causation. More often it is purely one of policy, of our more or less inadequately expressed ideas of what justice demands, or of administrative possibility and convenience, none of which have any connection with questions of causation at all.” (Prosser, op. cit. supra, pp. 240-41). [48 N.J. at 77-78.]
The Caputzal and Goldberg rationales are the same. Both seek in the first instance a determination as to whether a duty exists. Caputzal simply adds that some courts have expressed the issue in the confusing language of causation and foreseeability. In both Goldberg and Caputzal, this Court found that no duty existed.
A sharp line should be drawn between the duty question discussed above and whether or not as a matter of law more than one reasonable conclusion might be drawn from the evidence. The trial court must decide whether the plaintiff has produced sufficient evidence from which the jury upon proper instructions may determine whether there is causal relationship in fact. This issue is always for the jury, except when the facts will support only one reasonable inference. Here plaintiff failed on both counts. He did not prove that defendant breached a duty owed to him. Even if there were a duty, he did not establish cause-in-fact.
I
Before we consider the question of duty, it is appropriate to bear in mind the salient facts. Defendant, a manufacturer, produced a residential gas space heating unit that was reasonably fit, safe, and suitable for its intended purpose, namely, to provide heat in a room. The manufacturer had incorporated appropriate safety devices in the unit. Plaintiff’s employer *178removed those safety devices. Approximately fifteen years thereafter plaintiff was burned when gas, which had allegedly accumulated due to the removal of the safety devices, ignited. Plaintiffs expert suggested that use of a left-handed thread nipple instead of a right-handed thread nipple would have made it more difficult to remove the safety devices.
We have generally adopted section 402A of the Restatement (Second) of Torts.1 Cepeda v. Cumberland Eng’g Co., 76 N.J. 152, 168, 169 (1978). That section reads as follows:
(1) One who sells any product in a defective condition unreasonably dangerous to the user or consumer or to his property is subject to liability for physical harm thereby caused to the ultimate user or consumer, or to his property, if
(a) the seller is engaged in the business of selling such a product, and
(b) it is expected to and does reach the user or consumer without substantial change in the condition in which it is sold.
(2) The rule stated in Subsection (1) applies although
(a) the seller has exercised all possible care in the preparation and sale of his product, and
(b) the user or consumer has not bought the product from or entered into any contractual relation with the seller.
When the manufacturer produces a product that is reasonably fit, suitable, and safe for its intended purposes and it reaches the user without substantial change, then the general rule, as stated in section 402A, is that the manufacturer is not responsible. When the seller “delivers the product in a safe condition, and subsequent mishandling or other causes made it harmful by the time it is consumed,” the seller is not responsible. Restatement § 402A comment g. The Restatement, however, notes a possible exception to this rule when the seller expects the product to be processed or otherwise substantially changed before it reaches the user or consumer. Id. caveat (2), *179comment p. This exception may occur when the manufacturer should have anticipated changes in the normal course of usage of the product. See id. comment p (and examples therein). When safety devices are removed from a product, however, it is appropriate to ask whether responsibility should be imposed on the intermediate party who made the change or on the manufacturer.
The focus of the inquiry should be whether the ultimate result is fair and equitable. See Caputzal v. The Lindsay Co., supra, 48 N.J. at 77; Goldberg v. Housing Auth. of Newark, supra, 38 N.J. at 583. In this case and under these circumstances should the particular hazard fall within the scope of protection afforded by the doctrine of strict liability? The ultimate question is whether the standard proposed by plaintiff to utilize a left-handed thread nipple should be adopted. If so, it would of course affect an entire product line. See Twerski, “Seizing the Middle Ground Between Rules and Standards in Design Defect Litigation: Advancing Directed Verdict Practice in the Law of Torts,” 57 N.Y.U.L.Rev. 521, 545-46 (1982). The public that buys heaters would have to pay more. Id. at 544-45. Should those purchasers, who presumably are located in many states, bear the costs of the improper action of plaintiff’s employer?
Plaintiff’s proposed standard would also have an adverse impact upon servicing and component interchangeability. The uncontradicted evidence is that service people carry standard right-handed threads utilized in appliances made by other manufacturers. Further, the safety valve used by defendant is made by many companies and these valves are interchangeable. Thus, consumers’ heaters may now be serviced readily and expeditiously by various appliance dealers. Obtaining a special part from the factory would be more costly and time-consuming. Although technologically known, the left-handed thread unit has not been available in the industry.' Plaintiff’s proposed design change would substantially affect the utility in the field of defendant’s safety valve and the product.
*180What was the likelihood that the space heater would cause injury? Defendant had no records of any complaints concerning removal of the safety devices, though it had sold 50,000 residential space heaters over a twenty-year period. Throughout fifteen years of the misuse of the heater in question, the instant situation was apparently the only one in which injury occurred.
I question the fairness of condemning defendant’s design of its space heater product line predicated on this one incident when it appears that the design decision to install the safety devices made the heater reasonably fit, suitable, and safe for its intended purpose as a residential space heater. Defendant has provided an adequate safety device whose only alleged flaw is that it can be removed. Though there may be circumstances in which this is enough to constitute a defect (for example, if the safety device was so easy to remove that a small child could and did remove it), this is not such a situation.
Most cases that relate to the removal of safety devices when such removal is not required for the machine’s operation and maintenance have held that the manufacturer is not subject to strict liability. See Hanlon v. Cyril Bath Co., 541 F.2d 343, 345-46 (3d Cir.1975) (under Pennsylvania law, manufacturer entitled to a directed verdict because of “a substantial change,” as used in section 402A, when purchaser substituted a “significantly different and much more sensitive starting mechanism” for the original mechanism that protected against accidental activation); McGrath v. Wallace Murray Corp., 496 F.2d 299, 302-03 (10th Cir.1974) (under Utah law, misuse was a valid defense when purchaser’s employees chose to remove a guard on a grinder because the guard made their work “more difficult”); Tuttle v. U.S. Slicing Mach. Co., 335 F.2d 63 (4th Cir.1964) (meat grinder sold with protective grill held in place with screws “not an inherently dangerous instrumentality”); Ellison v. Northwest Eng’g Co., 533 F.Supp. 482 (S.D.Fla.1982) (summary judgment granted to manufacturer when affidavits established that guard was bolted in place at time machine was *181manufactured and delivered to purchaser, although the guard may subsequently have been removed); Baines v. U.S. Pipe & Foundry Co., 463 F.Supp. 107 (N.D.Ala.1979) (summary judgment granted to manufacturer both because product was not defective and no proximate cause existed since purchaser failed to assemble work deck with safety chains that were supplied); DeArmond v. Hoover Ball & Bearing Uniloy Div., 86 Ill.App. 3d 1066, 42 Ill.Dec. 193, 408 N.E.2d 771 (1980) (affirming summary judgment for manufacturer when safety doors were removed by purchaser, regardless of how easy it was to remove the doors); Coleman v. Verson Allsteel Press Co., 64 Ill.App. 3d 974, 21 Ill.Dec. 742, 382 N.E.2d 36 (1978) (affirming judgment for manufacturer because of substantial change when purchaser substituted easily-activated pedestal-mounted control panel for more difficult to activate shoulder-high controls); Robinson v. Reed-Prentice Div. of Package Mach. Co., 49 N. Y. 2d 471, 477, 479, 403 N.E.2d 440, 442-44, 426 N.Y.2d 717, 719-22 (1980) (no liability of manufacturer when safety gate and interlocks were so altered by purchaser as to be ineffective even though “manufacturer knew precisely” that the purchaser would so modify the machine); Temple v. Wean United, Inc., 50 Ohio St.2d 317, 364 N.E.2d 267 (1977) (manufacturer not liable because no defect existed at manufacture and purchaser altered an existing safety precaution by shifting activating buttons from shoulder to waist height); Ford Motor Co. v. Eads, 224 Tenn. 473, 480-82, 457 S.W.2d 28, 31-32 (1970) (no liability for manufacturer because “hot wiring” of tractor was an efficient intervening cause when the hot-wiring circumvented the safety mechanism designed to prevent tractor from being started in gear); cf. DeRosa v. Remington Arms Co., 509 F.Supp. 762, 769 (E.D.N.Y.1981) (citing Robinson, supra, for the proposition that “deliberate bypassing of safety devices [on a shotgun] for no reason associated with the user’s need to utilize the product effectively need not be guarded against” by the manufacturer). See generally Note, “Product Modification: The Effect of Foreseeability,” 42 U.Pitt.L.Rev. 431, 455 (1981) *182(to allow liability for foreseeable changes “effectively requires a foolproof design and makes the manufacturer an insurer”).
Our New Jersey case law is not to the contrary. While Cepeda v. Cumberland Eng’g Co., supra, 76 N.J. 152, involved removal of a safety device by someone other than the manufacturer, there the design defect was that the intended operation of the “pelletizing” machine required removal of the safety device. The Court in Cepeda found that
there was credible evidence that the guard had to be removed frequently during operations. Human nature being what it is, we believe a jury might have inferred that a reasonable manufacturer, from that fact, could have objectively expected that on occasion production with the machine would be resumed without replacing a guard just removed in a clean-out operation. [Id. at 180-81.]
The removal of the safety device from the gas heater in the instant case was not required in order to operate the gas heater for the purpose for which it was sold: heating rooms by a permanent installation. The holding of Cepeda is in line with earlier New Jersey case law that restricted the reach of strict liability to normal uses of a product and ordinary results of such uses. See Caputzal v. The Lindsay. Co., supra, 48 N.J. at 76 (defendant not liable for plaintiffs heart attack caused by fear because such a heart attack was an “extraordinary occurrence, not reasonably to be expected in a normal person”); Maiorino v. Weco Prods. Co., 45 N.J. 570, 574 (1965) (“A manufacturer or seller is entitled to expect a normal use of his product. The reach of the doctrine of strict liability in tort in favor of the consumer should not be extended so as to negate that expectation.”).
In Suter v. San Angelo Foundry & Mach. Co., 81 N.J. 150, 159-60 (1979), we acknowledged that misuse of a product — that is, usage for other than an intended or reasonably anticipated purpose — would not give rise to strict liability. We referred approvingly to nonliability in situations in which a person used a knife as a toothpick and a person contrary to directions walked on a leg supported by a pin that failed because of its misuse. Id. at 159-60.
*183The majority today has effectively reversed the position expressed in Suter. To state, as the majority does, that simply because an act is “foreseeable,” there is an automatic triggering of a jury question on whether the manufacturer should have guarded against misuse begs the question. Almost any human action is foreseeable. Under the majority’s rationale it is “objectively foreseeable” that someone might use a knife as a toothpick or a pin supporting a leg contrary to explicit instructions and that such misuses may therefore become jury questions.
The majority adds that the foreseeability must be “objective” or “reasonable.” Ante at 165-171. This seeming restriction, however, amounts to little in practical effect. In discussing foreseeability in the context of negligence, Dean Prosser has succinctly stated the problems inherent in reasonable foreseeability: “In one sense, almost nothing is entirely unforeseeable, since there is a very slight mathematical chance, recognizable in advance, that even the most freakish accident which is possible will occur, particularly if it has ever happened in history before.” W. Prosser, Law of Torts § 43 (4th ed. 1971) (emphasis added, footnote omitted).
The removal of a safety device is always objectively foreseeable in the sense that someone, somewhere, is bound to remove it. What product could not have been designed so that a particular injury would have been avoided? Utilizing this open-ended approach is not what strict liability is all about and attempting to limit it by words like foreseeability is nothing more than circular. See Caputzal v. The Lindsay Co., supra, 48 N.J. at 75 (“anything is foreseeable”); Goldberg v. Housing Auth. of Newark, supra, 38 N.J. at 583 (“Everyone can foresee the commission of crime virtually anywhere and at any time. If foreseeability itself gave rise to a duty to provide ‘police’ protection for others, every residential curtilage, every shop, every store, every manufacturing plant would have to be patrolled by the private arms of the owner.”).
*184II
Causal relation is a factual issue. To recover, plaintiff had the burden of proving that defendant improperly designed the space heater so that it was defective when placed in the stream of commerce and that the defect was a cause-in-fact of the accident and plaintiff's injuries. See Francis v. United Jersey Bank, 87 N.J. 15, 39 (1981). Plaintiffs theory was that defendant should have used a left-handed thread nipple instead of a right-handed thread nipple in the internal piping, and that if that had been done, the accident would probably not have occurred. Plaintiff, however, produced no evidence from which one could reasonably infer or conclude that plaintiffs employer would not have removed the safety devices had the space heater utilized a left-handed thread or that those in the construction industry generally who were misusing the heaters would not have continued to do so by removing safety devices. Plaintiff simply showed that it would have been somewhat more difficult to remove the controls — not that the removal would probably not have occurred. While proof of certainty is not required, the evidence must generate an inference of probability as distinguished from mere possibility. Cf. Hansen v. Eagle-Picher Lead Co., 8 N.J. 133, 141 (1951) (upholding defendant’s motion for judgment when proofs failed to establish that defendant’s conduct probably caused accident when plaintiff was struck on head by piece of metal that fell from building under construction). It was plaintiff’s burden to establish that the alternative design he proposed would have prevented the accident. This he failed to do.
An analogous example has been posed by Professor Twerski. A kitchen blender that broke when a metal spoon was placed in it should have been designed with one-quarter inch shatterproof glass. The injured user seeks recovery from the manufacturer, asserting defective design. Professor Twerski observes that “[i]f the injury would have happened even with the better design, then causation has not been established.” Twerski, “The Many Faces of Misuse: An Inquiry Into the Emerging *185Doctrine of Comparative Causation,” 29 Mercer L.Rev. 403, 419-20 (1978).
Plaintiff has not met his burden of proof of causation. Plaintiff at best proved that an alternative design would probably have made it more difficult to remove the safety devices. There was no showing that the increased difficulty would probably have prevented plaintiffs employer from removing the safety devices. Accordingly, plaintiff did not meet his burden of proving that the alleged defect was the cause-in-fact of the accident and his injuries.
I agree with the majority that the judgment of the Appellate Division should be reversed and the trial court’s judgment of dismissal reinstated.
Justice CLIFFORD and Justice GARIBALDI join in this opinion.
CLIFFORD, SCHREIBER and GARIBALDI, JJ., concurring in the result.
For reversal — Chief Justice WILENTZ and Justice CLIFFORD, SCHREIBER, HANDLER, POLLOCK, O’HERN and GARIBALDI — 7.
For affirmance — None.
We have substituted the phrase “reasonably fit, suitable and safe” for "unreasonably dangerous.” Suter v. San Angelo Foundry & Mach. Co., 81 N.J. 150, 176 (1979). We have differed with subsection (1)(b) of the Restatement in that the manufacturer remains liable even though he contemplates that the purchaser will add needed safety devices to the product. See Finnegan v. Havir Mfg. Corp., 60 N.J. 413, 423-24 (1972); Bexiga v. Havir Mfg. Corp., 60 N.J. 402, 410 (1972).