In 1978, Thomas Black, a fireman of twenty years experience, hosted a backyard barbecue. Most of the guests were experienced firefighters. Black, along with one Crockett, the then Fire Chief for the Town of Palm Beach, with eighteen years of firefighting experience, began the charcoal fire in mid-afternoon by saturating the briquettes with Sparky charcoal lighter fluid, a product manufactured by the appellant Husky Industries, Inc., and purchased that morning from a Winn-Dixie Store. Crockett lit the briquettes, and he and Black returned to the house.
About thirty minutes later, one of the guests noticed that the fire did not appear to be lit. Black retrieved the can of charcoal lighter. Seeing no flame, but not checking to see if the coals were still hot, Black reapplied fluid to the coals. A vapor cloud of lighter fluid immediately arose from the briquettes.
Knowing when he saw the vapor that the flash point of the combustible charcoal lighter had been reached and that the lighter fluid could explode at any moment, Crockett called out to Black to “move out of the way.” Black, having seen the vapor, also realized the coals were hot and backed up. Neither of them thought to douse the fire and begin again. Instead, Crockett waited approximately forty-five seconds and unsuccessfully tried to relight the fire from an air vent underneath the grill. Crockett again warned Black to back off and threw a match into the grill. The vapor ignited explosively. Crockett turned around and was walking away from the grill with his back to Black when he heard a second explosion. When Crockett turned around, Black, still holding the can of charcoal lighter, was on fire.
Both Black and Crockett were aware of the warning printed on the side of the Sparky can and knew that applying charcoal lighter to hot coals was contrary to that warning. It was undisputed that the manner in which the plaintiff and Crockett started the fire was in derogation of the warning and unsafe.
Black sued Husky and Winn-Dixie Stores for damages for the personal injuries suffered by him.1 He maintained that he was entitled to recover against the defendants on theories of implied warranty of merchantability and strict liability.2 The defects in the Sparky can which he alleged existed were that the can closure did not have a flashback arrester and the color of the can was black. At appropriate junctures in the trial the defendants moved for directed verdicts, asserting that Black’s proof failed to show the existence of these defects.3 The jury returned a verdict for *991Black and his wife in the amounts of $20,-000 and $1,000, respectively, but assessed Black’s negligence as a seventy-five per cent contributing cause of his injuries. The trial court accordingly entered judgment for Black and his wife in an appropriately-reduced amount. The defendants’ motion for judgment notwithstanding the verdict, again asserting the lack of proof of defect, was denied, and this appeal ensued.
The appellants’ primary contention on appeal is that the proof adduced by the plaintiffs was insufficient to present a jury question whether the product, by reason of its design (that is, the failure to provide for a flashback arrester in the can closure or the can being colored black) was defective. Since we agree with this contention, we need not address the appellants’ other points on appeal.
This case involves alleged design defects.4 Sparky charcoal lighter fluid cans are designed to be black in color and without flashback arresters. The plaintiffs’ burden was to show that this design was defective. Ford Motor Co. v. Hill, 404 So.2d 1049 (Fla.1981). In analyzing whether this burden has been satisfied, we must keep in mind two related and well-accepted propositions. First, a manufacturer is not an insurer, Houdaille Industries, Inc. v. Edwards, 374 So.2d 490 (Fla.1979); West v. Caterpillar Tractor Co., 336 So.2d 80 (Fla.1976); and second, a manufacturer does not have to make a product accident proof, Rice v. Walker, 359 So.2d 891 (Fla. 3d DCA 1978); Royal v. Black & Decker Manufacturing Co., 205 So.2d 307 (Fla. 3d DCA 1978); Builders Shoring & Scaffolding Equipment Co. v. Schmidt, 411 So.2d 1004 (Fla. 5th DCA 1982). Accord, Hunt v. Blasius, 74 Ill.2d 203, 384 N.E.2d 368, 372 (1978):
“A manufacturer is not under a duty in strict liability to design a product which is totally incapable of injuring those who foreseeably come in contact with the product. Products liability does not make the manufacturer an insurer of all foreseeable accidents which involve its product. Virtually any product is capable of producing injury when put to certain uses or misuses....
“[T]he availability of an alternative design does not translate into a legal duty in products liability. An action is not maintainable in products liability merely because the design used was not the safest possible.”
A defectively designed product is one that has been negligently designed. As noted by Professor Prosser:
“[In certain areas] the liability of the manufacturer, even though it may occasionally be called strict, appears to rest primarily upon a departure from proper standards of care, so that the tort is essentially a matter of negligence.
“One of these [areas] involves the design of the product, which includes plan, structure, choice of materials, and specifications. There is no doubt whatever that the manufacturer is under a duty to use reasonable care to design a product that is reasonably safe for its intended use and for other uses which are foreseeably probable. The question turns on what is reasonable care and what is reasonable safety. The maker is not required to design the best possible product, or one as good as others make, or a better product than the one he has, so long as it is reasonably safe. But the fact that others are making similar products with a safer design may be important evidence bearing upon the defendant’s reasonable care. Likewise the fact that others make use of the same design is evidence for the defendant, although it is not always conclusive.” Prosser, The Law of Torts 644—45 (4th ed. 1978).
The Flame Arrester
Black called two witnesses to testify about the purported significance of the absence of the flame arrester. The first was *992Robert Bean, the Fire Chief at the Trail Park Fire Department. Bean, apparently knowledgeable in the general field of fire prevention safety, including the storage of flammable and combustible materials, testified that his training included information about devices used in containers to prevent flashback.5 Bean had no experience with arresters, much less experience with arres-ters used in containers of the size and type involved in the present case. Bean was not aware whether other companies making charcoal lighter fluid equipped their cans with arresters. While Bean had examined the method by which vapors are vented for gasoline tanks, he had never “paid a flashback arrester any particular attention.” He admitted that the National Fire Code and all other codes he had examined required arresters on containers of flammable liquid only and acknowledged that charcoal lighter fluid is not a flammable liquid.6 Finally, and most significantly, Bean could not say whether the presence of a flashback arres-ter in the Sparky can would have prevented the accident in this case.
Bean’s testimony was clearly insufficient to prove that the absence of a flashback arrester in the Sparky can was a defect. First, it is apodictic that expert testimony is not admissible at all unless the witness has expertise in the area in which his opinion is sought. See Kelly v. Kinsey, 362 So.2d 402 (Fla. 1st DCA 1978). Bean admitted that the area of his expertise did not include the closing or capping of containers of the size in question in this case and that, indeed, his only familiarity with arresters related to flammable, not combustible, materials.
Second, even assuming, arguendo, that the defendants’ objection to Bean’s qualifications to testify was properly overruled, Bean’s opinion (“I think that the cap itself appears to be somewhat of a problem”) was, nevertheless, based on data relating solely to flammable materials in containers totally dissimilar in size to the one involved in this case. It has always been the rule that an expert opinion is inadmissible where it is apparent that the opinion is based on insufficient data. See Martin v. Story, 97 So.2d 343 (Fla. 2d DCA 1957) (opinion of public safety department expert that towed car was a dangerous instrumentality inadmissible where basis for opinion was admittedly incomplete statistics, and expert had no knowledge of the vehicle under discussion). See also Southern Utilities Co. v. Murdock, 99 Fla. 1086, 128 So. 430 (1930); Farley v. State, 324 So.2d 662 (Fla. 4th DCA 1975). Thus, in Cosgrove v. Estate of Delves, 35 A.D.2d 730, 315 N.Y.S.2d 369 (N.Y.App.Div.1970), a judgment for the plaintiff, who suffered an eye injury when the cork of a champagne bottle popped out when the foil covering the stopper was unwrapped, was reversed because of a failure to prove a defect. The plaintiff’s expert testified that in his opinion the design of the stopper was such that it could not resist the usual amount of pressure of a properly chilled bottle of champagne. The court, holding the opinion for naught, said:
“We are of the belief that the expert’s opinion was against the weight of the evidence in view of his failure to subject *993this or any other similar stopper to testing under pressure. In the absence of actual testing, the expert’s opinion is insufficient in the face of the widespread use of such plastic stoppers in the industry.” 315 N.Y.S.2d at 370 (emphasis supplied).7
The rule precluding expert testimony based on insufficient data, although procedurally modified by Florida’s evidence code so as to allow the giving of the opinion without prior disclosure of the underlying facts or data, remains substantively the same. Thus, Section 90.705, Florida Statutes (1979), provides in pertinent part:
“If the party establishes prima facie evidence that the expert does not have a sufficient basis for his opinion, the opinions and inferences of the expert are inadmissible unless the party offering the testimony establishes the underlying facts or data.”8
Moreover, not only must the underlying facts or data form a sufficient basis for an expert’s opinion, but the underlying facts or data upon which the opinion is based must themselves be relevant. Although the explicit relevancy requirement of Section 90.-702, Florida Statutes (1979), i.e., “the [expert] opinion is admissible only if it can be applied to evidence at trial,” might arguably be read to mean that the opinion is admissible simply if it relates to a fact in issue, we think the better reading of the rule is that “in order for the opinion to be admissible it must [in addition] be based on relevant facts.” Ehrhardt, 5 Florida Evidence 702.1, n. 16. But whether Bean’s testimony was non-probative because based on irrelevant as well as insufficient data, it was nonetheless non-probative.
Third, even taking Bean’s opinion (“I think the cap itself appears to be somewhat of a problem”) at face value, it falls woefully short of the proof required to demonstrate a design defect, because, as is all too well known, “[t]he judgment of an expert must be more than a guess.” Southern Utilities Co. v. Murdock, 99 Fla. at 1091, 128 So. at 432.
Finally, Bean’s admission that he could not say whether the presence of an arrester would have made any difference in this case totally negates the evidentiary value, if any, of his opinion that the “cap itself appears to be somewhat of a problem.”
The plaintiffs’ other witness concerning the flame arrester was Robert Westmoreland, a sales representative for the Continental White Cap Division, the company which manufactured the cap used on the Sparky can. Westmoreland testified that Continental had tested the cap with a flame arrester, and the result of the testing was that “we cannot prove that it does work and we cannot prove that it doesn’t work.”9 Needless to say, this testimony hardly advanced the plaintiffs’ cause and quite clearly was insufficient to prove a defect.
The Color of The Can
Chief Bean’s testimony was no more helpful to the plaintiffs in proving that the color of the can was a defect than it was in proving that the absence of a flashback arrester was a defect. Bean conceded that he was not an expert in the packaging of charcoal lighter fluid, or in colors “as far as *994what they are allowed to be.” When asked to state an opinion as to what the outside color of vessels containing flammable or combustible liquids should be, Bean responded: “I am not sure that I can answer that except to say that informally I think I know what they are.” He then stated that his “opinion” is “that I believe it is the accepted practice to paint them a light color such as silver or white.”
Bean admitted that he had consulted no authorities dealing with the coloration of cans or color absorption. While he said that everyone knows that black gets hot faster, and related a story about a dissimilar black tank coated with tar exploding when installed underground, he admitted to being totally incapable of addressing the effect that the material of the surface of the vessel would have on his opinion, although he believed it would make a difference.
For the very same reasons that we rejected Bean’s testimony as being non-probative on the flashback arrester issue, we reject it here. Bean was not qualified to render an opinion on the issue, since he had no expertise in the area in which his opinion was sought. Moreover, even assuming his qualifications, the opinion rendered was based on insufficient and dissimilar data so as to be irrelevant to show that the black color of the can was a design defect. Finally, Bean’s belief that it is the accepted practice to paint vessels containing flammable or combustible liquids a light color tells us nothing from which even an inference can be drawn that a can made of a particular metal, containing combustible charcoal lighter, is negligently designed, that is, defective, when painted black.
We turn now to the testimony of Gerald Fishe, a registered professional engineer and certified general contractor, called by the plaintiffs as an expert in the field of solar radiant energy and heat transfer. In Fishe’s opinion, the contributing factors which led to the explosion were that (1) the fluid remaining in the can was above the flash point of that fluid, (2) the wind at the particular time must have been still, (3) the vapor in the can above the liquid lighter fluid was explosive, and (4) the cause of the ignition was a vapor trail from the can to the grill.
Fishe’s testimony that the fluid in the can was above the flash point was, in turn, based on his conclusion that the black color of the can absorbed a lot of sun at a rapid rate. He arrived at that conclusion in two ways. First, he consulted and relied upon a manual which dealt with the percentage rate at which various colors absorb or reflect the sun. The experiment underlying this study was, according to Fishe, done by placing different colored containers of a like or similar liquid (presumably, although Fishe did not so state, a flammable or combustible liquid) on a picnic table in direct sunlight. Fishe admitted, however, that the most that the experiment showed was “a general concept or comparison,” that is, the comparative absorption rate of different colors. The experiment as described in the manual did not take into account the nature of the surface (e.g., dull or shiny finish).10 Moreover, Fishe admitted that he could not estimate how many British Thermal Units the Sparky can in the present case had absorbed, since any answer would depend on how full the can was and the can’s position in relation to the sun as it sat on the table, variables unanswered by the facts presented in this case. Fishe stated that he himself conducted no experiment with a like Sparky can of charcoal lighter fluid, because there were too many variables, and if the variables were to be eliminated, the cost of such an experiment would be prohibitive.
The second source of Fishe’s opinion was a manual which showed how quickly the temperature of a black roof would rise. The manual showed that the surface temperature of a black roof could rapidly rise to 160 degrees.
We think that Fishe’s testimony was appropriately characterized by defense counsel when he asked:
*995“[W]hat you have done is, you have read certain data and research and looked at the can and you have sat down and thought about this subject and you have come up with certain ideas in your head as to how this may have happened.”
It is apparent that had Fishe not merely relied on the manuals describing these two experiments, but had himself conducted the experiments, the results of the experiments would have been inadmissible because of the dissimilarity of circumstances and conditions. See Hisler v. State, 52 Fla. 30, 42 So. 969 (1906); State v. Arroyo, 422 So.2d 50 (Fla. 3d DCA 1982); Vitt v. Ryder Truck Rentals, Inc., 340 So.2d 962 (Fla. 3d DCA 1976). See also Burgin v. Merritt, 311 So.2d 688 (Fla. 3d DCA 1975). It should be no less apparent that since the basis for Fishe’s opinion that the fluid in the Sparky can had reached its flash point was experiments that themselves were inadmissible because non-probative, that the opinion itself was rendered worthless. And to the extent that Fishe arrived at his opinion independent of the experiments in the manual, it was, as he candidly but unintentionally admitted, based on pure speculation and guesswork as to the amount of fluid in the Sparky can and the can’s position in relation to the sun. As defense counsel appropriately asked, “Sir, how do you solve all of these unknowns in your head that you cannot solve in the practical world by putting them together in an experiment?”
But even taking Fishe’s testimony at face value, it still is insufficient to show that the black color of the Sparky can was a defect. At best, the testimony shows (a) the can was black, (b) black heats up faster than other colors, (c) black heats to a high temperature fast, (d) the liquid in the can had reached its flash point, and (e) the liquid being at its flash point was a substantial contributing factor to the accident. What is missing here is any showing through Fishe or anyone else that if the can had been any other color or made of any other material, it would not have exploded after forty-five minutes in the hot sun. There is simply nothing from which a trier of fact could infer that the color of the can was a defect.11
Having reviewed the evidence in a light most favorable to the plaintiffs as we are required to do, see Rodi v. Florida Greyhound Lines, 62 So.2d 355 (Fla.1952); Gatto v. Publix Supermarket, Inc., 387 So.2d 377 (Fla. 3d DCA 1980), we are nonetheless impelled to conclude that the record in this case is devoid of any evidence from which a jury of reasonable persons could find that the Sparky can of charcoal lighter fluid was defective, either in not having a flashback arrester or by virtue of the black color of the can. Accordingly, we find that the trial court erred in denying the appellants’ motions for directed verdict and for judgment notwithstanding the verdict.
REVERSED AND REMANDED, with directions to enter judgment for the appellants.
DOWNEY and GLICKSTEIN, JJ., concur.