MALLETT, C.J.

We granted leave in these consolidated cases to determine whether the Court of Appeals properly adopted the "manifestation trigger theory" to deny coverage under standard comprehensive general liability and umbrella excess insurance policies allegedly issued by the defendant insurers. The policies essentially provided coverage for an "occurrence" taking place within the policy period. Plaintiffs sought coverage for expenses relating to environmental clean-up actions where the contamination was discovered years after the policy periods had expired, but where it allegedly occurred while the policies were in effect. The Court of Appeals held that coverage under the policies was not triggered because the contamination was not discovered within the policy periods, i.e., it did not manifest within the policies' effective dates. Because we find no support for the "manifestation trigger" in the relevant policy language, we reverse.

I FACTS AND PROCEEDINGS A GELMAN SCIENCES

Plaintiff Gelman Sciences, Inc., manufactures microporous filters in a plant near Ann Arbor, Michigan. In 1964, it began using 1,4-dioxane on an experimental basis in its manufacturing process. From 1966 to 1984, Gelman regularly used 1,4-dioxane as a solvent to dissolve a polymer used in manufacturing filter sheets. Rinse water from the manufacturing process contained trace amounts of 1,4-dioxane. To dispose of this rinse water, Gelman utilized, pursuant to permit, a system of wastewater treatment ponds designed to biodegrade organic wastes. For two years in the late 1960s, treated water from one of the ponds leaked into a marshy area behind plaintiff's facility. Plaintiff consulted the Water Resource Commission and, pursuant to its suggestion, stopped the overflow by dredging out the bottom of the pond to allow the wastewater to seep into the ground. In the 1970s, Gelman, also pursuant to permit, began disposing of wastewater using a spray irrigation system. Gelman was allegedly unaware that these permitted waste systems would not biodegrade 1,4-dioxane. In 1985, the Washtenaw County Health Department discovered that certain drinking water wells located near plaintiff's facility were allegedly contaminated with 1,4-dioxane. The state and numerous private parties sued Gelman for damages relating to the groundwater contamination.

Plaintiff Gelman, for itself and on behalf of one of its insurers with whom it had settled, brought this action, seeking to establish that defendants owed a duty to indemnify and defend these lawsuits under various standard comprehensive general liability (CGL) and umbrella excess insurance policies allegedly in effect between the years 1963 and 1969. Although Gelman has not located the actual policies, it has listed the policy numbers of the various CGL and umbrella policies and quotes the language relating to coverage contained in the standard documents.

The trial court granted summary disposition for the insurers, finding that even if the policies in fact existed, coverage was not triggered because the groundwater contamination was not discovered until after they expired. The Court of Appeals affirmed on the trigger of coverage issue. 214 Mich. App. 560; 543 N.W.2d 38 (1995). We granted leave to appeal, limited to the issue whether the lower courts erroneously applied the date of manifestation of injury as the appropriate trigger of coverage for determination of insurance carrier responsibility to the insured.

B ARCO INDUSTRIES

This Court recounted the following summary of the factual and procedural background in Arco Industries Corp. v. American Motorists Ins. Co., 448 Mich. 395, 399-401; 531 N.W.2d 168 (1995), when it was previously before us concerning a different issue:

Plaintiff Arco Industries Corporation is a small automotive parts manufacturer that has operated a manufacturing plant in Schoolcraft, Michigan, since 1967. As part of the manufacturing process, the automotive parts are dipped into liquid plastisol or vinyl. Volatile organic compounds (VOCs) such as perchloroethylene, trichloroethylene, 1-2 dichloroethylene and vinyl chloride, were used to clean the parts during the manufacturing process and to remove plastisol from the plant floors. The plant floor was designed with a trench drain system that drained waste from the plant floor into an unlined seepage lagoon located in the back of the plant. As a result, VOCs contaminated the seepage lagoon and ground water.

In November, 1985, the Department of Natural Resources notified Arco that the seepage lagoon was contaminated with VOCs, and records indicated that Arco was the source of the contamination. After Arco's failure to resolve the problem, the DNR filed suit against Arco in federal court in an attempt to compel Arco to remedy the VOC contamination and collect claimed response costs. Subsequently, the State of Michigan and Arco entered into a consent decree whereby Arco agreed to pay the state $450,000 in response costs together with attorney fees. Arco also agreed to develop and implement a multimillion dollar ground water and soil remediation program.

Arco's insurer, AMICO, refused to defend or indemnify in the underlying litigation alleging that the insurance contract did not cover this type of incident. As a result, on February 4, 1987, Arco filed suit against AMICO, seeking to compel the insurer to honor its contractual obligations. In response to the suit, AMICO's defense was that this type of incident was not a covered "occurrence" within the meaning of the applicable comprehensive liability policies because Arco either expected or intended the pollution that resulted from its manufacturing process.

The trial court found that the contamination was not anticipated by Arco, and that there was no "showing that there was an intention by anyone to contaminate." Thus, on September 28, 1990, a judgment was entered compelling AMICO to pay its allocated share (68.63 percent) of all indemnifiable losses up to the aggregate limits of AMICO's coverage of $3.5 miliion.

The Court of Appeals, however, reversed the trial court's decision and held that AMICO did not have the responsibility to defend or indemnify Arco. The Court held that there were clearly intentional discharges of VOCs by Arco employees and Arco either should have foreseen the result of the intentional acts, knew, or should have known that such practices would result in a substantial probability that VOCs would contaminate the soil and ground water. 198 Mich. App. [347] 352-353 [ 497 N.W.2d 190 (1993)].

This Court reversed and remanded, holding that a subjective, rather than an objective, standard applies to determining whether the insured expected or intended the harm. Arco, supra. On remand, the Court of Appeals again held that the defendants owed no duty of coverage, this time relying on the "manifestation trigger" theory to hold that because the contamination was not discovered during the policy periods, coverage had not been triggered. 215 Mich. App. 633; 546 N.W.2d 709 (1996). This Court granted leave on the trigger of coverage issue and consolidated the case with Gelman Sciences.

II BACKGROUND

Comprehensive general liability policies insure against liabilities to third parties. The coverage typically is broad, extending generally to any business liability not expressly excluded. Since the early 1940s, the insurance industry has drafted and periodically revised the standard CGL policy. In the 1960s, it modified the standard CGL policy from insuring against "accidents" to insuring against "occurrences." The standard CGL policies involved in these cases are primarily of this latter variety. They provide coverage for an "occurrence," defined typically as "an accident, including injurious exposure to conditions, which results, during the policy period, in bodily injury or property damage, neither expected nor intended from the standpoint of the insured."

Courts have typically had difficulty applying the standard CGL language in cases involving gradual, continuous, or delayed property damage from pollution. The apparent difficulty is in determining when coverage is triggered under the relevant policies when the actual damage is not discovered until years after the pollution began.

Courts and commentators have discussed four possible theories for determining what event or events trigger coverage under standard CGL policies. These are the "exposure," "injury in fact," "manifestation," and "continuous" trigger theories. See, e.g., Ray Industries, Inc. v. Liberty Mut. Ins. Co., 974 F.2d 754, 764-765 (CA 6, 1992). These various theories are relevant to the question when an occurrence takes place for purposes of determining if there is coverage under a particular insurance policy.

The exposure trigger places the "occurrence" at the earliest time, i.e., when the exposure to injury-causing conditions occurs. In the environmental pollution context, this is the date on which the pollution began. The manifestation theory places the "occurrence" at the latest possible time, i.e., when the property damage is ultimately discovered. Some courts describe this trigger as being when the injury manifests itself in the form of ascertainable property damage. The trigger under the injury-in-fact approach is actual property damage. Under this theory, the finder of fact must determine when exposure to the pollutants resulted in actual property damage in the facts of a given case. The final theory, known alternatively as the "continuous," "triple," or "multiple" trigger, holds that the injury occurs continuously from exposure until manifestation. Thus, all insurers who assumed risk from the initial pollution through discovery of property damage would be liable.

A

In the cases we review today, the Court of Appeals applied the manifestation theory to determine that the insurers owed no coverage. Before explaining why we are convinced that the manifestation approach is error, we will examine the Court of Appeals analysis.

In Gelman, the Court agreed with, and followed, the analysis in Mraz v. Canadian Universal Ins. Co., Ltd, 804 F.2d 1325 (CA 4, 1986). In Mraz, the policyholders buried barrels containing chemical wastes in 1969. Contaminated soil and water were not discovered at the burial site until 1982. The policyholders sought coverage for losses relating to the contamination. The policy in Mraz contained language similar to the policies at issue: it covered losses for an "occurrence," which was defined as "an accident, including injurious exposure to conditions, which results, during the policy period, in bodily injury or property damage, neither expected nor intended from the standpoint of the insured." The Court quoted at length from Mraz:

"The general rule is that `[t]he time of the occurrence of an accident within the meaning of an indemnity policy is not the time the wrongful act was committed but the time when the complaining party was actually damaged.' United States Fidelity Guaranty Co. v. American Ins. Co., 169 Ind. App. 1 [7]; 345 N.E.2d 267, 270 (1976). Often, these cases involve a wrongful act that produces no harm for a period of time and then suddenly manifests itself in a burst of damage. See, e.g., Travelers Ins. Co. v. CJ Gayfer's Co., 366 So.2d 1199, 1202 (Fla.App., 1979) (faulty roof drainage system finally leaks); Singsaas v. Diederich, 238 N.W.2d 878, 880 (Minn., 1976) (collapse of negligently installed manlift); Peerless Ins. Co. v. Clough, 105 N.H. 76 [78-79]; 193 A.2d 444 446 (1963) (fire in negligently built chimney).

"There are situations, however, in which the existence or scope of damage remains concealed or uncertain for a period of time even though damage is occurring. The leakage of hazardous wastes as in this case is a clear example. Determining exactly when damage begins can be difficult, if not impossible. In such cases we believe that the better rule is that the occurrence is deemed to take place when the injuries first manifest themselves. See Appalachian Ins. Co. v. Liberty Mutual Ins. Co., 676 F.2d 56, 62 (CA 3, 1982); cf. Bartholomew v. Appalachian Ins. Co., 655 F.2d 27, 28 (CA 1, 1981) (when the defect takes place or is discovered); Aetna Casualty Surety Co. v. PPG Industries, Inc., 554 F. Supp. 290, 294 (D Ariz., 1983) (coverage is based upon the time the damage was discovered).

"Therefore, we hold that in hazardous waste burial cases such as this one, the occurrence is judged by the time at which the leakage and damage are first discovered." [ 214 Mich. App. 567-568, quoting Mraz, supra at 1328.]

The Court of Appeals also stated that the trial court, in applying the manifestation trigger, had properly relied on Transamerica Ins. Co. v. Safeco Ins. Co., 189 Mich. App. 55; 472 N.W.2d 5 (1991). The policyholder in Transamerica installed urea-formaldehyde insulation in businesses and homes and sought coverage for losses relating to personal injuries sustained from exposure to harmful gases released by the insulation. The Court of Appeals in Gelman identified the issue in Transamerica as whether persons and property exposed to harmful gases were injured, for purposes of triggering coverage, when the insulation was installed or when the symptoms manifested themselves. The Court then stated that Transamerica held that coverage was triggered when the personal injury or property damage resulting from exposure manifested itself to the party claiming injury.

The Court also acknowledged cases from other jurisdictions rejecting the manifestation approach and distinguished them on the basis that in those cases, the environmental damage could be pinpointed either to a single discharge or to an exact time frame. Ray Industries v. Liberty Mut. Ins. Co., supra, Upjohn Co. v. New Hampshire Ins. Co., 178 Mich. App. 706; 444 N.W.2d 813 (1989), rev'd on other grounds 438 Mich. 197; 476 N.W.2d 392 (1991), Inland Waters Pollution Control, Inc. v. Nat'l Union Fire Ins. Co., 997 F.2d 172, 179-189 (CA 6, 1993).

The Court of Appeals in Arco simply cited Gelman and Transamerica, without providing any further analysis, before applying the manifestation trigger to deny coverage.

III LAW AND ANALYSIS

While the various trigger theories are useful to describe the outcomes reached by courts in different factual settings, we think that reference to trigger theories can be deceiving. Ultimately, it is the policy language as applied to the specific facts in a given case that determines coverage. In this regard we agree with the observations made by United States District Judge Churchill:

Comparison of the manifestation theory with the injury in fact theory is absolutely meaningless unless there is some real possibility of a substantial time lag between the actual injury and the resulting manifestation. Similarly, comparison of the injury in fact theory with the exposure theory means nothing unless the advocate of the exposure theory presses for a ruling that coverage was triggered prior to the point when injury in fact occurred. Moreover, in all of these scenarios, "real" or "actual" injury must be defined in temporal relation to initial exposure or ultimate manifestation. Any effort to logically organize all of the trigger cases along these lines would be perforce ineffectual. Consequently, trigger rulings are most appropriately derived by reference to the operative policy language, as opposed to the judicial gloss placed upon similar language in ostensibly analogous cases. [ Dow Chemical Co. v. Associate Indemnity Corp., 724 F. Supp. 474, 479 (ED Mich., 1989).]

We find Judge Churchill's comments particularly insightful. In reviewing several trigger cases from other jurisdictions, we think it apparent that some courts too quickly label the "trigger theory" they are supposedly applying without carefully considering the factual distinctions, or lack thereof, between exposure, injury in fact, and manifestation. Even more troubling, some courts are too hasty in applying a particular trigger without carefully considering the relevant policy language. We must not forget that the issue presented today is fundamentally a question of insurance contract interpretation.

A

Well-settled principles govern our interpretation of the CGL policies at issue. We must first determine whether the policy is clear and unambiguous on its face. We must not create an ambiguity where none exists. Further, we may not rewrite the plain and unambiguous language under the guise of interpretation. Instead, we must enforce the terms of the contract as written, interpreting the unambiguous language in its plain and easily understood sense. Upjohn, 438 Mich. 206-207.

Additionally, in order to protect the policyholder from confusing policy language, ambiguous provisions must be construed in favor of the policyholder. Powers v. Detroit Automobile Inter-Ins. Exchange, 427 Mich. 602, 623; 398 N.W.2d 411 (1986); American Bumper Mfg Co. v. Hartford Fire Ins. Co., 452 Mich. 440, 448; 550 N.W.2d 475 (1996). Further, where the policyholder's interpretation in favor of coverage is reasonable, that reasonable expectation must be enforced. Dow Chemical Co., supra. "Under the rule of reasonable expectation, this Court will examine whether `the policyholder, upon reading the contract language is led to a reasonable expectation of coverage.'" Vanguard Ins. Co. v. Clarke, 438 Mich. 463, 472; 475 N.W.2d 48 (1991), quoting Powers, supra at 632.

Turning next to the express policy language, Gelman asserts that defendants' policies contained the following language concerning coverage:

The company will pay on behalf of the Insured all sums which the Insured shall become legally obligated to pay as damages because of

A. bodily injury or

B. property damage

to which this Insurance applies, caused by an occurrence, and the company shall have the right and duty to defend any suit against the Insured seeking damages on account of such bodily injury or property damage, even if any of the allegations of the suit are groundless, false or fraudulent, and may make such investigation and settlement of any claim or suit as it deems expedient, but the company shall not be obligated to pay any claim or judgment or to defend any suit after the applicable limit of the company's liability has been exhausted by payment of judgments or settlements.

[A]n accident, including injurious exposure to conditions, which results during the policy period, in bodily injury or property damage neither expected nor intended from the standpoint of the insured. . . . [Emphasis added.

Arco's policies issued by defendant American Motorists Insurance Company contain virtually identical language.

The plain language of the policies at issue unambiguously provides that an "occurrence" is an accident that results in property damage during the policy period. Otherwise stated, according to the policies' explicit terms, actual injury must occur during the time the policy is in effect in order to be indemnifiable, i.e., the policies dictate an injury-in-fact approach. The manifestation trigger simply is not supported by the policy language. Consequently, we hold that the Court of Appeals clearly erred in adopting and applying the manifestation trigger to deny coverage.

B

We are not alone in concluding that the standard CGL policy language concerning when coverage is triggered is unambiguous. Other courts, some of which have attempted to discern how this Court would rule on the issue, have determined that the language explicitly supports an "injury in fact" approach. For example, the United States District Court for the Northern District of Ohio, in a case applying Michigan law, determined as follows:

This court also concludes that the "occurrence" clause is not ambiguous. This court believes that faced with the question, the Michigan Supreme Court would adopt the injury in fact trigger of coverage. . . . [ Detrex Chemical Industries, Inc. v. Employers Ins. of Wausau, 746 F. Supp. 1310, 1324-1325 (ND Ohio, 1990).]

In adopting the injury-in-fact trigger, the court in Detrex reviewed decisions from other jurisdictions that are helpful to further explain operation of the injury-in-fact trigger in cases involving continuous, gradual, or delayed property damage. The court cited with approval the following language from Triangle Publications, Inc. v. Liberty Mut. Ins. Co., 703 F. Supp. 367, 370 (ED Pa, 1989):

The plain meaning of the term "occurrence," as used in the CGL policy, is clear. It is (1) an accident (2) which results (3) in property damage (4) during the policy period. Stated another way, an actual injury must occur during the time the policy is in effect in order to be indemnifiable or compensable. Any other interpretation is simply not supported by the CGL policy language.

We agree. We also agree with the following analysis, again cited with approval in Detrex, from Abex Corp. v. Maryland Casualty Co., 252 U.S. App. DC 297, 305; 790 F.2d 119 (1986):

The plain language of the definition of "occurrence" used in the CGL policy requires exposure that "results, during the policy period, in bodily injury" in order for an insurer to be obligated to indemnify the insured. The unambiguous meaning of these words is that an injury — and not mere exposure — must result during the policy period. The CGL policies expressly distinguish exposure from injury; to equate the two as urged by Abex is to ignore this distinction. Any argument that mere exposure — without injury — triggers liability is simply unsound linguistically. Additionally, the manifestation theory, too, is inconsistent with the definition of "occurrence." Although the language of these policies demands that the insured prove that an exposure caused an injury during the policy period, it imposes no requirement that the injury be discovered at that time.

Another court, also attempting to apply Michigan law, has determined that the standard CGL policy language unambiguously dictates an injury-in-fact trigger.

Although the concept expressed by these carefully selected words [in the standard CGL policy] may be difficult to apply to some claims made against the policies, the meaning of the operative language is not ambiguous.

Absolutely nothing in the policy language suggests that an event can trigger coverage prior to the time that the event results in property damage. By the same token, nothing suggests that exposure to conditions triggers coverage prior to the time that property damage results from such exposure. Moreover, the policy language does not even hint that property damage must be known to anyone in order to trigger coverage. Likewise, nothing in the policy language indicates that property damage does not exist unless someone knows about it. [ Dow Chemical Co., supra at 481.]

C

Both Gelman and Arco strenuously argue that the Court of Appeals application of the manifestation trigger effectively converts their more costly "occurrence" policies into less expensive "claims made" policies because coverage would be denied unless the property damage were discovered within the policy period. We find merit in this argument. In St. Paul Fire Marine Ins. Co. v. American Home Assurance Co., 444 Mich. 560, 569, n 19; 514 N.W.2d 113 (1994), we noted that, under an occurrence policy, "[b]oth insured and insurer anticipate that the claim would probably not be made until after the policy period and possibly when another policy is in effect." This Court has also noted:

Coverage in an "occurrence" policy is provided no matter when the claim is made, subject, of course, to contractual and statutory notice and limitations of actions provisions, providing the act complained of occurred during the policy period. Because the insurer's liability in such policies ordinarily relates to a definite, easily identifiable and notorious event such as an automobile accident, . . . the insurer is ordinarily able to conduct a prompt investigation of the incident . . . with the result that actuarial considerations permit relative certainty in estimating loss ratios, establishing reserves, and fixing premium rates.

"Claims made," or "discovery" policies, on the other hand, are of relatively recent origin and were developed primarily to deal with situations in which the error, omission, or negligent act is difficult to pinpoint and may have occurred over an extended period of time. [ Stine v. Continental Casualty Co., 419 Mich. 89, 98-99; 349 N.W.2d 127 (1984).]

From the insurer's standpoint, in retrospect, property damage resulting from gradual or latent contamination would have been better suited to a claims-made policy. This does not change the fact, however, that the policies at issue are "occurrence" policies and does not provide a basis to deny coverage by application of a judicially created manifestation theory. See Marathon Flint Oil Co. v. American States Ins. Co., 810 F. Supp. 850, 852-853 (ED Mich., 1992).

D

Like other courts that have adopted the manifestation trigger, the Court of Appeals in Arco and Gelman justified the manifestation trigger because of its perception that determining the precise timing of actual property damage would be difficult, if not impossible. While we appreciate the difficulty of proof in this regard, this difficulty cannot justify redrafting unambiguous policy terms in the guise of judicial interpretation.

Instead, courts should endeavor to determine which policies are triggered from the evidence presented. Where a plaintiff can show that property damage occurred sometime within one or several of the relevant policy periods, and the plaintiff presents credible evidence (such as expert testimony) that fairly supports the plaintiff's claims regarding when property damage occurred, courts should accept such evidence as dispositive.

Admittedly, in some cases in which the plaintiff has been able to establish that property damage occurred sometime within one or several policy periods, the plaintiff may not be able to pinpoint when the property damage actually occurred, or to apportion the amount of damage occurring during each period. In such cases, we agree with those commentators that suggest that courts should establish some means of allocating the risk among the insurers. While we do not decide the best method of allocation today because we did not grant leave on that issue and it has not been adequately briefed and argued, we note that other courts have utilized various proration methods and joint and several liability to allocate liability. See, e.g., Ins. Co. of North America v. Forty-Eight Insulations, Inc., 633 F.2d 1212, 1224-1225 (CA 6, 1980); Keene Corp. v. Ins. Co. of North America, 215 U.S. App. DC 156; 667 F.2d 1034 (1981).

In advocating a rule of pro-rata allocation among all insurers sharing the risk, one commentator has noted:

In the absence of such a rule, the insured will be denied at least some of its CGL indemnification, even though it has paid yearly premiums. In a way, each of the other three triggers — manifestation, exposure, and continuous — applies an injury-in-fact analysis, but with different definitions of "injury." The exposure theory deems the injury to have occurred when the victim or property is exposed to the damage-causing agent. The manifestation trigger deems the injury to have occurred when it is discovered. Thus, these other triggers all simplify the fact-finding involved by taking certain shortcuts in determining when the injury occurred. A court applying the injury-in-fact trigger eventually will reach an outcome resembling that which would be produced by one of the other three triggers (or perhaps some variation thereof). The advantage of the injury-in-fact trigger is that it produces a more accurate, less arbitrary result. . . . Thus, the position that apportionment is unacceptable because it relies on factual assumptions that are unfair to the defendant would invalidate all trigger theories except injury-in-fact. This position is not logically consistent if it rejects [proration], and yet supports the manifestation, exposure, or continuous trigger. Those triggers also make factual assumptions that can not be justified anymore [sic] than [proration].

On the other hand, it is logically defensible to argue for a strict injury-in-fact trigger, with no allowance for the insured's inability to prove when damages or injuries occurred. Such a position, however, leads to unnecessarily harsh results. The insured would receive no CGL indemnification, even though the injury or damage occurred sometime during the policies covered by the insurers in the litigation. For these insurers to benefit from the insured's inability to prove the exact time of a causal fact would enrich them unjustly and would upset the reasonable expectations of the insured. [Comment: Nailing Jello to a wall: A uniform approach for adjudicating insurance coverage disputes in products liability cases with delayed manifestation injuries and damages, 83 Cal. L R 1243, 1290-1291 (1995).]

We agree with this reasoning insofar as courts should not employ a strict standard of proof regarding injury in fact and should instead endeavor to fairly allocate the risk.

E

In addition to relying on the convenience provided by the manifestation approach, the Court of Appeals in both Gelman and Arco also relied on two cases, Transamerica and Mraz, supra. We find, however, that neither decision justifies the Court of Appeals departure from the policies plain language.

As explained earlier, Transamerica involved bodily injury and property damage claims arising from exposure to gases emitted by urea-formaldehyde foam insulation (UFFI). Gas emissions were continuous from installation until removal, but were highest immediately after installation. The trial court found that the bodily injury and property damage alleged by the underlying plaintiffs occurred contemporaneously with installation. The Court of Appeals rejected the insured's argument that exposure alone triggers coverage. Instead, the Court held that the policy language dictated that "it is bodily injury or property damage . . . that triggers coverage." We find that the Court's statement of the proper trigger up to this point in its opinion is correct. The Court, however, went on to state that

this does not fully answer the question presented, which is: When does bodily injury or property damage occur where exposure to gases from UFFI is claimed? While there is little dispute that the effects of UFFI gases continue until the insulation is removed from the home and that the underlying plaintiffs alleged continuing injuries or damages, we find that bodily injury, for purposes of triggering coverage, occurs when the symptoms begin or manifest themselves to the homeowner. If this is at the time of installation, then plaintiff would be solely responsible for its insured's defense costs and damages. . . . [ Transamerica at 59.]

We find that the Court in Transamerica, while starting down the right track in its analysis, went astray in labeling manifestation as the trigger. To the extent that the language in Transamerica advocates a manifestation trigger, it is erroneous and should not be followed.

We further find that the other case relied on by the Court, Mraz, supra, is not persuasive authority and should not have been followed. The Court relied on Mraz for its conclusion that a manifestation trigger is appropriate because of the difficulty in determining exactly when damage begins. As we have already stated, a theory adopted for its convenience, which goes directly against the plain policy language, must be rejected. Further, as we have also already explained, other means may be utilized by courts to deal with the difficulties in proof that often exist in these types of cases.

Additionally, we note that Mraz, which was a federal district court decision interpreting Maryland law, has not been followed by Maryland courts. In Harford Co. v. Harford Mut. Ins. Co., 327 Md. 418; 610 A.2d 286 (1992), Maryland's highest court rejected the manifestation approach in favor of an injury-in-fact trigger. While agreeing with the United States Court of Appeals for the Fourth Circuit in Mraz that difficulty in proof is a problem in these types of cases, the court in Harford did not regard that difficulty to be justification for abandoning the plain language of the policies and noted that the factual determination was "quite likely a matter for expert testimony." Id. at 436. We agree and find that the Court of Appeals erred in its reliance on Mraz.

IV APPLICATION OF AN INJURY-IN-FACT APPROACH

In Gelman, coverage under its alleged policies is triggered when the alleged property damage, here groundwater contamination from Gelman's disposal of 1,4-dioxane, first occurred. Coverage would also exist under any subsequent policy periods during which contamination continued to occur. Because proceedings in the trial court had not been concluded before the trigger issue was appealed, we remand the case to the trial court for further proceedings consistent with this opinion. Particularly, the trial court must determine whether the policies in fact existed and when the contamination occurred.

In Arco, the trial court found that there had been coverage triggering occurrences in each of defendant's policy periods. Specifically, the court found as follows:

It was clear that there were accidents occurring throughout the period of the policies at issue. There were accidental spills, according to the testimony, which resulted in discharges into the drains in the plant, as well as unintentional overflows into the drains in the plant. The evidence supports finding that the VOCs which were released, entered the soil in all three areas described on the exhibits, that they entered the aquifer and contributed to the plume.

These findings were based on testimony from plaintiff's employees regarding numerous and repeated accidental spills and overflows of materials containing VOCs. Additionally, Arco established, by expert testimony, that not only were there releases of contaminants in each policy period, but also that there was actual property damage resulting from these releases in each of the policy periods as well. Joel Hunt, an expert on hydrogeology and contaminant transport, testified that the releases to the soil would have reached the groundwater beneath the site in eight weeks or less and would have contributed to the soil and groundwater contamination. The trial court found Mr. Hunt's testimony to be credible. We affirm the trial court's conclusion that the evidence supported finding that the VOCs were released and contributed to the contamination in each of the policy periods.

V CONCLUSION

The standard CGL policy language, providing coverage for property damage occurring during the policy period, unambiguously dictates application of an injury-in-fact trigger of coverage. The Court of Appeals clearly erred in applying a manifestation trigger to hold that there was no coverage under the policies because the environmental contamination was not discovered during the policy periods. Consequently, we reverse the Court of Appeals decisions in both Gelman and Arco. In Gelman, we remand to the trial court for proceedings consistent with this opinion. In Arco, we affirm the trial court's findings that there had been coverage triggering occurrences in each of the defendant's policy periods.

BRICKLEY, CAVANAGH, BOYLE, WEAVER, KELLY, and TAYLOR, JJ., concurred with MALLETT, C.J.