Davidson, J.,

delivered the opinion of the Court.

Early in 1971, Michelle Cadeaux, the appellee, did what thousands of Americans do every year — she bought a car. Unlike many of her compatriots, however, she was so unhappy with her purchase that within six months she stopped driving the car and in another year she was in court suing the appellant, Lustine Chevrolet, for breach of contract, breach of warranty, and fraud. She was seeking $15,000 compensatory damages and $25,000 punitive damages.

On 13 and 14 September 1972, in the Circuit Court for Prince George’s county, the case was tried by a jury presided over by Judge Robert B. Mathias. Judge Mathias granted directed verdicts for appellant on the counts involving breach of contract and breach of warranty and as to the punitive damages sought in relation to the fraud count. The issue of fraud was submitted to the jury which returned a verdict in favor of the appellee in the amount of $4,000 compensatory damages.

*32 The facts are not in dispute. In late January 1971, appellee went to appellant’s showroom where she spotted a Camaro which she liked very much. A salesman informed her that the car was a 1970 “demonstrator” which had been driven 5,100 miles by Mr. Lustine’s grandson and carried a new-car warranty. She specifically asked if the car previously “had been wrecked” and was told by the salesman that it had not. Had she been told that the car had been in an accident, she would not have bought it, but because of the salesman’s assurance, she purchased the car.

It was then that her troubles began. When appellee attempted to drive the Camaro out of appellant’s driveway, it moved about one foot, stalled and then would not start. Appellant installed some new plugs and points and appellee drove home. About four days later, appellee was driving on an interstate highway in Virginia at a speed of approximately 70 miles per hour when the car stalled right in the middle of the road, leaving the power steering and power brakes totally inoperative. It was towed back to Lustine where it was discovered that “something electrical” had been wrong and the car was repaired. On another occasion, when she was driving on the Baltimore-Washington Parkway in a pouring rainstorm, the windshield wipers became inoperative. Being unable to see, appellee nearly collided with another car as she attempted to get her Camaro off the road. She drove the car home, but the next morning it would not start and had to be towed to Lustine where it was repaired. On still another occasion, the car stalled in a “bad section of Washington,” started up again, but then stalled by the time appellee had driven into Virginia. Once again, Lustine had to tow the car back and repair it. In addition to these dramatic episodes, there were at least four other occasions during the first two months of her ownership when appellee was unable to start the car in her own driveway and Lustine had to tow it away and make repairs.

As a result of her unpleasant experiences, appellee became fearful of driving her car and after April drove it less and less frequently. By July 1971, appellee gave possession of her car to her brother who had just returned from Viet Nam. By *33 this time she was so fearful of driving that she gave up her driver’s license. During the six months that appellee drove the car, it was never involved in an accident. Nonetheless various difficulties other, than those previously described were encountered. She complained to Lustine that when the car was driven at 60 miles an hour it would shake, “pull to the right,” and “almost jump into the next lane,” that the windshield wipers “banked,” and that the radio did not work properly. The battery and left front headlight were replaced at some point. By June of 1971, the stalling ceased to be a problem. The car was aligned and the wheels were balanced, but it continued to shake when driven at high speeds. The appellee was never charged by Lustine for either towing or repairs.

Mr. Charles Jourdak, appellee’s brother, testified that after July 1971, the car went through tires and “shocks” quickly. He took it to various repair shops to have the alignment checked and the wheels balanced, but the car never operated properly. He testified that throughout the period of his possession, the car was never in an accident.

In December 1971, Mr. Jourdak noticed that body putty was falling out of the fender of the car. In the belief that the car might have been in an accident before his sister bought it, he had the car examined by Mr. Gilbert A. Bell, Jr., an auto damage appraiser, who qualified at trial as an expert in the field of estimating auto body damage. Mr. Bell inspected the car on 21 March 1972 when the odometer read 33,921 miles. Mr. Bell found that the car had been in a collision whose impact was centralized on the left front fender. He testified that the impact causing the collision was “not too great.” Sheet metal repairs and body repairs had been made to the front left fender. He further found that the car had a loose and very rough suspension, which he termed “spongy.” The springs were not strong enough to hold up the front end of the car, and the motor mounts were weakened or cracked. A rubber bumper which normally acts to protect the frame from the suspension by hitting the frame when the car drops suddenly or hits a “heavy hole” was striking the frame of the car every two or three times the car jolted, which made the *34 car noisy. Mr. Bell attempted to give an opinion as to whether the car’s troubles were attributable to the previous accident, but his answer was stricken because it dealt in possibilities rather than probabilities.

The fact that the car had been in an accident prior to its purchase by appellee was confirmed by Mr. Frank Katen, the Lustine corporation’s treasurer. He testified that he had learned from the insurance report filed after the accident that it had occurred on 20 October 1970, when a car came parallel to the Camaro and hit its left front fender. The repairs were made by Lustine and their accident repair sheet indicated that repairs were necessary to the bumper, the bumper bracket, the radiator support, the front fender, and some molding. The car was balanced at that time. The total cost of the repairs was $149.22.

Appellant contends that the trial court erred in denying its motion for a directed verdict on the question of fraud. It maintains that appellee failed to prove the essential elements of fraud. More particularly, it asserts that the appellee failed to prove that the misrepresentation was made for the purposes of defrauding her; that she suffered any compensable damage; and that any damage which she may have suffered resulted directly from the misrepresentation. Appellee contends that she established all of the essential elements of fraud. More specifically, she asserts that she showed that but for appellant’s misrepresentation as to the car’s previous involvement in an accident, she would not have purchased the car; that the misrepresentation was made to induce her to purchase the car; that she suffered actual damage in the form of nervousness and fear which impaired her physical and mental health; and that such damage resulted from the fraud which induced her to purchase the car. For reasons set forth below, we find that the evidence is insufficient to establish that any damage which may have been suffered by appellee resulted directly from a misrepresentation of fact. Accordingly, the other contentions need not be considered.

It is well settled in Maryland that in order to recover in an action for fraud, the plaintiff must show that: 1) the *35 representation made was false; 2) its falsity was either known to the speaker, or the misrepresentation was made with such a reckless indifference to truth as to be equivalent to actual knowledge; 3) it was made for the purpose of defrauding the person claiming to be injured thereby; 4) such person not only relied upon the misrepresentation but had a right to rely upon it in the full belief of its truth, and would not have done the thing from which the injury resulted had such misrepresentation not been made; and 5) he actually suffered damage directly resulting from such fraudulent misrepresentation. Appel v. Hupfield, 198 Md. 374, 378, 84 A. 2d 94, 95-96 (1951); Gittings v. Von Dorn, 136 Md. 10, 15-16, 109 A. 553, 554 (1920); see Walsh v. Edwards, 233 Md. 552, 558, 197 A. 2d 424, 427 (1964); Schmidt v. Millhauser, 212 Md. 585, 592, 130 A. 2d 572, 575 (1957); Babb v. Bolyard, 194 Md. 603, 609, 72 A. 2d 13, 16 (1950). “It is well settled by all the authorities that the fraud must work an actual injury on the party complaining . . . .” Reynolds v. Evans, 123 Md. 365, 367, 91 A. 564, 565 (1914); see Russell v. Stoops, 106 Md. 138, 143, 66 A. 698, 699-700 (1907); Weaver v. Shriver, 79 Md. 530, 545, 30 A. 189, 190 (1894). Recently the Court of Appeals in Empire Realty Company, Inc. v. Fleisher, 269 Md. 278, 284, 305 A. 2d 144, 147 (1973), explicated the concept of causal connection in fraud cases when it said:

“One suing for fraud or deceit must establish that he sustained damage by reason of the fraud, and that his injury was the natural and proximate consequence of his reliance on the fraudulent act.” (Citations omitted.)

Thus it is clear in Maryland that in order to recover for fraud, the plaintiff must show not only that he would not have performed the act from which the injury resulted but for the misrepresentation, but also that the fact misrepresented was the proximate cause of the injury. Translated to the facts of this case, this means that the plaintiff was required to show not only that she would not have purchased the car but for a misrepresentation as to its *36 previous involvement in an accident, but also that its previous involvement in an accident was the proximate cause of the car’s subsequent mechanical difficulties and malfunctions.

The requirement that a causal connection be shown between the defendant’s act and the plaintiff’s harm is the same in both the tort actions of fraud and negligence, Richardson v. Boato, 207 Md. 301, 304-06, 114 A. 2d 49, 51-52 (1955), and in our view, the general principles governing the mode of proof for this element is also the same for both. With respect to negligence cases, it has been held that in order to establish causal connection, the plaintiff must show that there is a reasonable probability or reasonable certainty that the act complained of caused the injury suffered. Mere possibility is not enough. Sun Cab Co. v. Carter, 14 Md. App. 395, 408, 287 A. 2d 73, 80 (1972). Reasonable probability can be shown by either direct evidence or inferences drawn from surrounding circumstances. Peterson v. Underwood, 258 Md. 9, 17, 264 A. 2d 851, 855 (1970). The necessary inferences may be drawn by laymen where the disability develops coincidently with or within a reasonable time after the act complained of, where the causal connection is clearly apparent in the circumstances themselves, or where the cause of the injury relates to matters of common experience, knowledge or observation of laymen. However, where the cause of the harm is a complicated question involving fact finding which properly falls within the province of an appropriate expert, proof of the cause must be made by such witnesses. Kraft v. Freedman, 15 Md. App. 187, 194, 289 A. 2d 614, 618, cert. denied, 266 Md. 736 (1972); see Wilhelm v. State Traffic Comm., 230 Md. 91, 99-100, 185 A. 2d 715, 719 (1962). This is equally true whether the expert is a physician, as in Kraft and Wilhelm, or a person trained in the evaluation of automobile damage, as in the case at bar. Thus, in the instant case, in order to recover, the plaintiff must have introduced some evidence to show a reasonable probability that the malfunctioning of the Camaro after her purchase was caused by the damage sustained by the vehicle in the accident of 20 October 1970.

*37 The record clearly establishes that the Camaro had been in an accident in October 1970; that prior to the purchase, appellant’s salesman had assured appellee that the car had not previously been wrecked; that such assurance was a misrepresentation of a fact; and that but for that misrepresentation of fact appellee would not have purchased the car. However, we find that there was insufficient evidence to show that the malfunctioning of the car subsequent to its purchase resulted from the damage it had sustained in the accident.

The record shows that the impact of the collision on 20 October 1970 was centralized on the left front fender and was “not too great.” Repairs were made to the bumper, the bumper bracket, the radiator support, the front fender and some molding, and the car was balanced. The total cost of repairs was $149.22. The record shows further that after appellee bought it, the car was plagued with various mechanical difficulties. It frequently would not start. It stalled even when being driven at high speeds on the highway. When driven at a speed above 60 miles an hour, the car shook and pulled to the right. Various repairs were made by Lustine and by June 1971, the stalling ceased. However, the car continued to shake when driven at high speeds, even after numerous attempts to remove this difficulty had been made. In March 1972, when the car had been driven some 33,000 miles, an examination revealed that it had a loose and very rough suspension; the springs were not strong enough to hold up the front end; the motor mounts were weakened or cracked; and a rubber bumper designed to protect the frame from the suspension was not operating properly. However there was no direct evidence presented to show a reasonable probability that the damage sustained by the car occasioned in the October 1970 collision was the proximate cause of the car’s subsequent mechanical difficulties and malfunctions. Nor was there any expert testimony to establish that fact with the requisite reasonable certainty.

Whether a collision of “not too great” impact, which resulted in repairs costing $149.22 to a bumper, a bumper *38 bracket, a radiator support, a front fender, and some molding, subsequently caused a car to fail to start, stall, •shake, pull to the right, and, some 28,000 miles later, have loose and rough suspension, inadequate springs, cracked motor mounts, and a defective rubber bumper is a complicated question. Under such circumstances, proof of a causal connection between the appellant’s misrepresentation and the malfunctioning of appellee’s car cannot be based solely upon inferences drawn by laymen. In the absence of evidence to show that the defects complained of did not exist prior to the accident, or subsequent to the repair of the vehicle but prior to its purchase, the fact that the car was purchased only two and a half months subsequent to its repair does not permit a rational inference that the malfunctions of the vehicle complained of by appellee were caused by the October 1970 collision. That the circumstances do not clearly give rise to an inferénce of causation or involve matters within the common knowledge of laymen is manifested by the fact that appellant’s own expert in auto body repairs was unable to state with reasonable certainty that the collision probably caused the subsequent malfunctioning of the automobile. Craig v. Chenoweth, 232 Md. 397, 401, 194 A. 2d 78, 79-80 (1963). To allow a jury of laymen to attempt to determine questions of the complexity here involved would permit the rankest kind of guesswork, speculation and conjecture. Worthington Constr. Co. v. Moore, 266 Md. 19, 28, 291 A. 2d 466, 470 (1972); Wilhelm v. State Traffic Comm., supra, 230 Md. at 101, 185 A. 2d at 719. We hold that under the circumstances of this case, the evidence of causal connection between the misrepresentation and the harm suffered was insufficient to submit the question of proximate cause to the jury. The trial court should have directed a verdict for appellant.

Judgment reversed.

Costs to be paid by appellee.