Justice Guzman delivered the opinion of the Court.
A word is not a crystal, transparent and unchanged, it is the skin of a living thought and may vary greatly in color and content according to the circumstances and the time in which it is used.
-Oliver Wendell Holmes
We have long articulated a principle of contract construction that permits courts to consult the facts and circumstances surrounding a negotiated contract's execution to aid the interpretation of its language.2 Despite expounding on this principle from time to time, and as recently as last term,3 it remains susceptible to confusion and inconsistency when applied to unambiguous contract terms. The principle's limitations are, however, clear: surrounding facts and circumstances cannot be employed to "make the language say what it unambiguously does not say"4 or "to show that the parties probably meant, or could have meant, something other than what their agreement stated."5 In other words, extrinsic evidence may only be used to aid the understanding of an unambiguous contract's language, not change it or "create ambiguity."6
When interpreting a written contract, the prime directive is to ascertain the parties' intent as expressed in the instrument. "[O]bjective, not subjective, intent controls,"7 so the focus is on the words the parties chose to memorialize their agreement. But language is nuanced, and meaning is often context driven.8 Contract language is thus construed in its lexical *758environment, which may include objectively determinable facts and circumstances that contextualize the parties' transaction.9 Surrounding facts and circumstances can inform the meaning of language10 but cannot be used to augment, alter, or contradict the terms of an unambiguous contract.11
This case involves construction of a settlement agreement that conditioned resumption of uranium mining operations on restoration of well-water quality if pre-mining data showed the water had been suitable for specified uses before prior mining operations began. Construing the contract, the lower courts held that, in determining whether a restoration obligation exists as to a disputed well, the mining company was contractually required to ignore data showing no pre-mining suitability. In so holding, the courts impermissibly employed surrounding facts and circumstances to divine subjective intent and interpolate constraints not found in the contract's unambiguous language. We therefore reverse the court of appeals' judgment and render judgment for the mining company.
I. Background
More than a decade ago, the Texas Commission on Environmental Quality (TCEQ) authorized URI, Inc. to commence uranium mining operations in the third production area (PAA 3) of the Kingsville Dome Mine in Kleberg County, Texas.12 Mining operations were suspended, however, due to opposition from the County and local activists, including a group called South Texas Opposes Pollution (STOP). Ensuing litigation between Kleberg County and URI was resolved in December 2004 after protracted and contentious negotiations produced a settlement agreement (Settlement Agreement). This lawsuit concerns URI's performance under that agreement.
In the proceedings below, the parties clashed over URI's compliance with myriad terms of the Settlement Agreement, and all matters were ultimately resolved favorably to URI except questions about its compliance with section 11.1 of the agreement. That provision, which is the only one at issue here, conditions URI's resumption of mining operations in PAA 3 on URI's certification that wells in a previously mined area-PAA 1-have been restored to pre-mining water quality if the water had been suitable for specified uses before URI began mining that area in 1988. Section 11.1 provides:
11. Other Provisions.
11.1. Resumption of PAA 3 Mining To Await PAA 1 Restoration Certificate and Data. URI will not resume economic mining in KVD [Kingsville Dome] Production Area 3 before it has submitted to the County Judge of Kleberg County both *759(1) a sworn statement of a URI officer certifying to the Kleberg County Judge
(i) that URI has completed treatment of 6 pore volumes of water in KVD Production Area 1; and
(ii) that 90% or more of the combined total of TCEQ production area baseline wells in KVD Production Area 1 and any other wells in the production patterns that URI sampled and for which baseline is available before mining begins, whose water was suitable on a well-by-well basis before URI's mining in PAA 1 for use as either drinking water, livestock water, or irrigation water according to the criteria of 1.7(3) and parameters and limits set out in the Water Quality Use Limits Table in Par. 1.7 of this agreement, has been returned to suitability, based on the same parameters and limits on a well-by-well basis, for the same use to which it was suited before such mining; and,
(2) a copy of the analytical data the URI officer relies upon in making the certification under this sub-par. 11.1 as to (i) which individual baseline wells are deemed to have produced water suitable for either drinking, stock watering, or irrigation use before URI's mining in PAA 1 and (ii) the use suitability to which post-mining water quality in the relevant Production Area 1 baseline wells has been returned.
In 2007, URI recommenced mining in PAA 3 after certifying that (1) 6 pore volumes of water in PAA 1 had been treated and (2) no wells in PAA 1 were subject to restoration to the levels specified in the parties' agreement. Kleberg County disputed URI's compliance with subsection 11.1(1)(ii)'s 90%-restoration requirement and resolved to enforce the Settlement Agreement before any further mining could take place.13
Advancing conflicting constructions of section 11.1, URI sought a declaration that it had complied with its contractual obligations while Kleberg County countersued for breach of contract and declaratory relief to the contrary. In addition to damages, Kleberg County sought abeyance of PAA 3 mining through specific performance of section 11.1. Though URI was undertaking restoration of PAA 1 wells in accordance with TCEQ mandates, the County argued the parties intended section 11.1 to impose a higher restoration standard before URI could proceed with further mining operations in PAA 3. The crux of the dispute is whether any of the PAA 1 wells described in subsection 11.1(1)(ii) had water suitable for drinking, livestock, or irrigation uses before URI started mining that area. If so, URI breached the Settlement Agreement when it began mining PAA 3 before restoring 90% of the subject wells to prior water quality.
No dispute exists regarding the combined total of wells subsection 11.1(1)(ii) encompasses that qualify as either (1) "TCEQ production area baseline wells in KVD Production Area 1" or (2) "any other wells in the production patterns that URI sampled and for which baseline is available before mining begins" in PAA 3. Sixteen wells fall within the former and twenty-three in the latter, for a total of thirty-nine *760wells with pre-mining baseline data. Nor is there any dispute that of those wells, thirty-eight were not subject to the restoration-to-suitability mandate in subsection 11.1(1)(ii) because they did not have water suitable for drinking, livestock, or irrigation purposes before URI commenced PAA 1 mining in 1988.
At issue is URI's contractual obligation to "return[ ] to suitability" a single well (Well I-11)14 whose water was suitable for irrigation purposes based on pre-mining baseline samples URI took in 1985 but not suitable for any of the contractually specified uses based on pre-mining baseline samples URI took in 1987. Averaging the two data points, URI determined the water in Well I-11 was not suitable for any of the contractually specified uses before URI began mining PAA 1 in 1988. Having undertaken the certification process outlined in the Settlement Agreement, URI took the position that it had no obligation under subsection 11.1(1)(ii) to "return[ ] to suitability" any of the PAA 1 wells before recommencing mining operations in PAA 3.
Relying on the 1985 baseline data, Kleberg County argued URI was obliged to restore Well I-11's groundwater to use for irrigation. According to the County, the Settlement Agreement prohibits use of the 1987 data because that information was not publicly available or otherwise made available to the County when the parties executed the Settlement Agreement, though it was subsequently filed with the TCEQ and accepted as "accurately reflect[ing] baseline quality."15 Kleberg County asserts the parties contemplated that only the 1985 data-which URI had submitted to the TCEQ as part of the application for the original PAA 1 mining permit-would be used to determine URI's well-restoration obligation and, based on that information, the parties intended that URI would restore Well I-11 to pre-mining suitability for irrigation uses.
As support for its construction of the Settlement Agreement, the County cites the 2004 transcript of a Kleberg County Commissioners' Court meeting which concluded with a majority of the commissioners voting to execute the Settlement Agreement. During the meeting, a member of STOP had interjected to express strong opposition to the proposed settlement's terms. Among other objections, STOP's representative argued section 11.1 was "totally worthless" because documentation she had obtained from TCEQ and provided to the commissioners showed only one well would be subject to restoration on the terms stated.16 The transcript *761indicates that some of the commissioners believed the Settlement Agreement's terms, separately and in the aggregate, would provide more affirmative relief than they could expect in the pending litigation and administrative proceedings. The County cites these circumstances as supporting a construction of the Settlement Agreement that prohibits use of the 1987 baseline data and requires restoration of Well I-11's water quality to suitability in accord with the 1985 baseline data.
Following a lengthy bench trial, the trial court found URI breached its obligation to restore Well I-11 to its pre-mining water use. With respect to section 11.1 of the Settlement Agreement, the trial court rendered judgment declaring:
1. URI, Inc., has completed the treatment of 6 pore volumes as required by paragraph 11.1(1)(i) of the Settlement Agreement;
2. The intent of the Settlement Agreement was for the I-11 well to be returned to the use for which it was suited pursuant to the 1985 sample, before URI, Inc. began mining in PA-3. The intent of the Settlement Agreement was for the Water Quality Use Limits table in paragraph 1.7 of the Settlement Agreement to control suitability determination making the I-11 well at the time the agreement was entered into suitable for agricultural irrigation.
3. URI, Inc., began mining PA-3, but did not restore Well I-11 to suitability for agricultural irrigation use and thereby unintentionally and without deliberate intent breached the Settlement Agreement.
4. A primary focus of the litigation was the restoration of Well I-11 to prior suitability. In order to allow for the restoration, and to allow URI, Inc. to continue to have the financial resources to complete the restoration, justice would not be served by prohibiting URI, Inc.'s continued mining of PA-3. The disagreement in the how [sic] the language of the contract was to be construed was a legitimate issue between the parties and the court does not find bad faith on the part of either party.
5. When URI, Inc. submitted the 1987 sample data for Well I-11 to TCEQ and sought to change the groundwater restoration requirements, even though the request complied with TCEQ procedures, the effect was an unintentional breach of the Settlement Agreement. The change altered the suitability of Well I-11 *762from suitable for agricultural irrigation to suitability for no purpose.
The trial court awarded Kleberg County $20 in nominal damages; ordered URI to continue restoring Well I-11 until the water was suitable for agricultural irrigation use; denied Kleberg County's request to bar URI from mining in PAA 3 until restoration of Well I-11 was completed; and declined to award attorney's fees to either party. In written Findings of Fact and Conclusions of Law, the court found the 1987 baseline data was valid but should not have been used to certify compliance with section 11.1 because "this additional data on Well I-11 was not public data and was not known to the County at the time of the execution of the Agreement; and thus, would not have been contemplated by the County at the time it entered the Agreement in 2004."
The parties cross-appealed on numerous points, the lion's share of which are not at issue here.17 The court of appeals affirmed as to all liability issues but reversed and remanded as to remedies, holding (1) Kleberg County is entitled to attorney's fees as the prevailing party on its claim that URI failed to comply with section 11.1 and (2) the trial court abused its discretion by excusing URI from complying with the Settlement Agreement's requirement that mining in PAA 3 abate until restoration of Well I-11 has been completed.18
Agreeing with the trial court's conclusion that URI breached section 11.1 by relying on the 1987 data, the court of appeals explained:
The plain language of paragraph 11.1 of the Agreement states that before URI resumes mining in PAA-3, URI must restore 90% of the sampled wells for which a pre-mining baseline is available that were suitable for either drinking water, livestock water, or irrigation before mining commenced to suitability for that use. It does not refer to any particular sample, or even a single sample. The problem with this approach is the necessary assumption that the chemical composition of the well (such as uranium) is static. However, according to the 1985 and 1987 sample disparities of the chemical composition, it is not. Therefore, we conclude that the Agreement allows for the use of only the 1985 well sample.
The trial testimony shows that the Commissioners adopted paragraph 11.1 in order to have the wells restored to a "higher level" than required by TCEQ. While the "higher level" is undefined, it is clear the Commissioners contemplated that the "higher level" would either be drinking water, livestock water, or irrigation water. Further, it is undisputed that the 1985 data showed that only one well-I-11-was suitable for any purpose. Since the agreement requires 90% of the wells in PAA-1 that were suitable for either drinking water, livestock water, or irrigation to be returned to that state prior to mining PAA-3, it necessarily contemplates that at least one well must be suitable.
Incorporating the 1987 sample data would mean that none of the wells in PAA-1 were suitable for any use, and consequently, any requirement for URI to restore wells to either drinking water, livestock water, or irrigation purposes before mining PAA-3 would be meaningless. Since incorporating the 1987 sample data would render the intended safeguards of paragraph 11.1 meaningless, *763we disagree with URI that the Agreement allows the use of the 1987 well sample data.19
On appeal to this Court, URI contends (1) the lower courts improperly considered extrinsic evidence of Kleberg County's subjective intent rather than construing section 11.1 according to its plain and unambiguous language and (2) according to the Settlement Agreement's plain language, the 1987 data may-and indeed, must-be considered in determining whether any of the PAA 1 wells with baseline data were suitable for any of the contractually specified uses "before URI's mining in PAA 1" began in 1988.
Kleberg County argues the agreement, informed by the facts and circumstances attending its execution, unambiguously establishes that the 1985 baseline data sets the restoration standard for Well I-11 because it was the only "baseline ... available before mining begins." In the alternative, Kleberg County asserts that, for purposes of section 11.1's restoration requirement, the 1987 data cannot be considered to the exclusion of or by averaging with the 1985 data to negate suitability established by the 1985 data. In other words, because the 1985 data shows Well I-11's "water was suitable ... before URI's mining in PAA 1 for use as ... irrigation water," URI was required to "return[ ] [it] to [that] suitability" regardless of whether the 1987 data-on its own or by averaging all available data-shows no suitability before PAA 1 mining commenced.
We take this opportunity to reaffirm that the "facts and circumstances" extant at the time a contract is executed may be consulted only to inform the meaning of the language the parties chose to effectuate their accord. In construing an unambiguous contract or in determining whether an ambiguity exists, courts may not seek the parties' intent beyond the meaning the contract language reasonably yields when construed in context.
II. Discussion
A. Contract Construction Principles
The parties rely on conflicting interpretations of section 11.1, but both assert the Settlement Agreement is unambiguous. A contract is not ambiguous merely because the parties disagree about its meaning and may be ambiguous even though the parties agree it is not.20 Both the presence of ambiguity and interpretation of an unambiguous contract are questions of law we review de novo using well-settled contract-construction principles.21
When a contract's meaning is disputed, our primary objective is to ascertain and give effect to the parties' intent as expressed in the instrument.22 Objective manifestations of intent control,23 not "what one side or the other alleges they *764intended to say but did not."24 We therefore "presume parties intend what the words of their contract say"25 and interpret contract language according to its "plain, ordinary, and generally accepted meaning" unless the instrument directs otherwise.26
We are also cognizant that "words are simply implements of communication, and imperfect ones at that. Oftentimes they cannot be assigned a rigid meaning, inherent in themselves. Rather, their meaning turns upon use, adaptation and context as they are employed to fit various and varying situations."27 Even a single word can carry subtle-and significant-differences in meaning when applied to different situations.
Accordingly, to home in on the meaning the parties intended, we have long allowed that words must be construed in the context in which they are used. Context is not, however, confined to the two-dimensional contractual environs in which the words exist but may also encompass "the circumstances present when the contract was entered."28 This is so because words are "the skin of a living thought,"29 and our quest is to determine, objectively, what an ordinary person using those words under the circumstances in which they are used would understand them to mean.
B. Surrounding Circumstances Only Inform Contract Language
The central issue here is the role evidence of "surrounding circumstances" plays in the construction of section 11.1 of the Settlement Agreement. Consideration of surrounding circumstances is limited by the parol evidence rule,30 which prohibits a party to an integrated written contract from presenting extrinsic evidence "for the purpose of creating an ambiguity or to give the contract a meaning different from that which its language imports."31 "Only where a contract is ambiguous may a court consider the parties' interpretation and 'admit extraneous evidence to determine *765the true meaning of the instrument.' "32 "[N]o issue regarding the parties' intentions is raised unless the [contract] is ambiguous-and evidence of those intentions cannot be used to create an ambiguity."33
The parol evidence rule does not, however, prohibit courts from considering extrinsic evidence of the facts and circumstances surrounding the contract's execution as "an aid in the construction of the contract's language,"34 but the evidence may only "give the words of a contract a meaning consistent with that to which they are reasonably susceptible, i.e. , to 'interpret' contractual terms."35 This is true even if doing so reveals a latent ambiguity in a contract's terms. But whether a court is considering if an ambiguity exists or construing the terms of an unambiguous contract, surrounding facts and circumstances can only provide context that elucidates the meaning of the words employed, and nothing else. As we have often stated in one way or another, "[u]nderstanding the context in which an agreement was made is essential in determining the parties' intent as expressed in the agreement , but it is the parties' expressed intent that the court must determine."36
If a written contract is so worded that it can be given a definite or certain legal meaning when so considered and as applied to the matter in dispute, then it is not ambiguous.37 But if contract language is susceptible to more than one reasonable interpretation when so viewed, an ambiguity exists.38
Contract ambiguity comes in two flavors: patent or latent.39 "A patent ambiguity is evident on the face of the contract" while a "latent ambiguity arises when a contract which is unambiguous on its face is applied to the subject matter with which it deals and an ambiguity appears by reason of some collateral matter," such as the circumstances present when the contract was entered.40 Allowing that surrounding circumstances may reveal a latent ambiguity seems conceptually at odds with the proscription against employing extrinsic evidence to create an ambiguity. But to say extrinsic evidence may not create an ambiguity but may reveal one means only that "[t]he ambiguity must become evident when the contract is read in context of the surrounding circumstances, not after parol evidence of intent is admitted to create an ambiguity."41
A classic example of a latent ambiguity is when a contract requires goods to be delivered to "the green house on Pecan Street," but there were, in fact, two green houses on Pecan Street.42 When surrounding circumstances reveal an ambiguity about the intent embodied in the contract's language, as in the "green house" example, *766extrinsic evidence of the parties' true intent will then-and only then-be admissible to settle the matter.43 But, when the contextual evidence discloses no ambiguity, extrinsic evidence that the parties actually intended for the goods to be delivered to the blue house on Pecan Street would not be admissible to alter unambiguous contract language requiring delivery to the green house.44 Nor would the contract's meaning be informed by extrinsic evidence that the parties intended additional requirements or constraints that were not expressed in the agreement-such as delivery by 5:00 p.m. or only on Sundays.
Two cases illustrate the interplay between latent-ambiguity analysis and a contract's surrounding circumstances. In Gallagher Headquarters Ranch Development, Ltd. v. City of San Antonio ,45 we held a latent ambiguity existed as to the matter in controversy. In that case, we considered the City of San Antonio's motion to dismiss the petition for review, which asserted a settlement agreement had mooted the controversy. The settlement agreement clearly settled two separate condemnation cases-one between Petitioner Gallagher and the City and another between Petitioner Chris Hill and the City.46 In the settlement agreement, Gallagher and Hill also purportedly agreed to release all claims "arising from or related to the events and transactions which are the subject matter of this cause."47 But the settlement agreement did not expressly identify the case before the Court, which involved both Gallagher and Hill along with a third-party (Petitioner Hooper) who was neither named in nor a signatory to the settlement agreement.48 Based on these circumstances, the Petitioners argued the settlement agreement did not encompass the appeal. Considering the release language and the existence of the pending appeal at the time the settlement agreement was executed, we held that "a latent ambiguity appear[ed] to exist" as it was "unclear whether the case [on appeal] ... is covered by the Agreement and release."49 We thus abated the petition and remanded the case to the trial court to prepare findings of fact on the issue.50
The extent to which parties may resort to surrounding circumstances to reveal a latent ambiguity is, however, limited, as demonstrated by National Union Fire Insurance Co. of Pittsburgh v. CBI Industries, Inc.51 In National Union , CBI Industries sought discovery in order to show a latent ambiguity in "absolute pollution exclusion" clauses used in their insurance policies and identical provisions throughout the industry.52 We held that discovery was unnecessary to interpret the policies' unequivocal language, because such discovery could only lead to improper parol evidence of subjective intentions.53
After an accidental explosion, CBI sought insurance relief from multiple insurers.54 The insurance companies moved for summary judgment on the ground that *767"absolute pollution exclusions" barred coverage.55 Two of the policies had nearly identical language excluding coverage for claims "caused by seepage or pollution or contamination of air ... however caused and whenever occurring."56 The trial court granted summary judgment, but the court of appeals held CBI was entitled to conduct discovery to raise a fact issue on latent ambiguity.57
We reversed the court of appeals, noting that the ambiguity issue is a question of law for the court and that the facts relating to the accident and the surrounding circumstances of the contract's execution were amply established.58 "Applying the policies' language to the context of the claim here does not produce an uncertain or ambiguous result, but leads to only one reasonable conclusion: ... coverage is precluded."59 Contrary evidence of industry intent was impermissible, we said, because "extrinsic evidence is inadmissible to contradict or vary the meaning of the [agreement's] explicit language."60
As Gallagher and National Union establish, the parol evidence rule prohibits extrinsic evidence of subjective intent that alters a contract's terms but "does not prohibit consideration of surrounding circumstances that inform, rather than vary from or contradict, the contract text."61 Thus, extrinsic evidence may be consulted to give meaning to the phrase "the green house on Pecan Street," but "cannot be used to show the parties' motives or intentions apart from" the language employed in the contract.62
Mindful of our responsibility "to honor the parties' agreement" without altering it,63 we have thus made a clear distinction between extrinsic evidence that illuminates contract language and extrinsic evidence that adds to, alters, or contradicts the contract's text. We recently added lucidity on this topic in First Bank v. Brumitt .64 Analogizing to the statutory-construction context, we explained what it means to consult " 'evidence' of 'attending circumstances' " to "aid in the construction of [a] contract's language" without running afoul of the parol evidence rule:
In the same way that dictionary definitions, other statutes, and court decisions may inform the common, ordinary meaning of a statute's unambiguous language, circumstances surrounding the formation of a contract may inform the meaning of a contract's unambiguous language. But courts may not rely on evidence of surrounding circumstances to make the language say what it unambiguously does not say.65
C. Facts And Circumstances That Aid Contract Interpretation
What "facts and circumstances" may be consulted will naturally vary from *768case to case, but reasonably well-defined contours can be mined from our jurisprudence. Because objective intent controls the inquiry,66 only circumstantial evidence that is objective in nature may be consulted. We have accordingly described surrounding circumstances as including " 'the commercial or other setting in which the contract was negotiated and other objectively determinable factors that give a context to the transaction between the parties.' "67 Setting can be critical to understanding contract language, as we found in cases involving the lawyer-client relationship68 and construction of an arbitration agreement.69 We have also cited trade custom as bearing on the parties' objective intent when provisions were stricken from a form insurance contract.70 Similarly, trade usage can illuminate the meaning of contract language because "the meaning to which a certain term or phrase is most reasonably susceptible is the one which [is] so regularly observed in place, vocation, trade or industry so 'as to justify an expectation that it will be observed with respect to a particular agreement.' "71 Facts attending the execution may or may not shed light on contract meaning and may or may not cross the parol-evidence line.72 In deciding what facts and circumstances are informative, rather than transformative, ascertaining objective meaning is the touchstone.
A certain degree of latitude is inherent in the inquiry, but absolute limits on the *769use of surrounding circumstances are abundantly clear. Parties cannot rely on extrinsic evidence to "give the contract a meaning different from that which its language imports,"73 "add to, alter, or contradict" the terms contained within the agreement itself,74 "make the language say what it unambiguously does not say,"75 or "show that the parties probably meant, or could have meant, something other than what their agreement stated."76
Here, the court of appeals impermissibly relied on extrinsic evidence of Kleberg County's subjective intent to construe the unambiguous language in section 11.1 of the Settlement Agreement. While evidence of the commercial setting necessarily contextualizes and informs the meaning of some of the contract terms (for example, "baseline wells," "baseline" data, and "before URI's mining in PAA 1"), the court of appeals went too far in looking beyond the Settlement Agreement's language to interlineate limitations and specific results not expressed in the instrument itself. Construing the contract in light of the surrounding circumstances here does not support the conclusion that section 11.1 precludes URI's reliance on the 1987 baseline data, and we may not rely on extrinsic evidence for the purpose of modifying, enlarging, or curtailing its terms.
D. The Contract Permits Consideration of 1987 Baseline Data
By its plain terms, section 11.1(1)(ii) imposes a well-restoration requirement contingent on an alteration of suitability that emanated from PAA 1 mining operations as determined by "baseline" data and the water quality that existed "before URI's mining in PAA 1." According to TCEQ regulations, "[b]aseline quality" refers to "[t]he parameters and their concentrations that describe the local groundwater quality of an aquifer prior to the beginning of injection operations."77 Section 11.1 does not refer to any particular set of baseline data, prescribe the use of particular data, place limitations on the data that may be considered in determining pre-mining suitability, or require restoration of any well, let alone a particular well.
The trial court found the 1987 baseline data URI relied on is valid, and the evidence at trial establishes that the TCEQ accepted the 1987 data as accurately reflecting baseline quality. Per the trial court's express findings, the 1987 baseline data shows the water in Well I-11 was unsuitable for any of the contractually specified uses before URI began mining in PAA 1 in 1988. The trial court's findings are unchallenged. Accordingly, under a plain reading of section 11.1(1)(ii), URI had no obligation to ensure the water in Well I-11 was suitable for drinking, livestock watering, or irrigation before resuming mining in PAA 3.
The lower courts concluded, however, that URI could not use the 1987 baseline data, alone or in combination with the 1985 baseline data, precisely because this outcome would ensue from a plain and objective reading of the Settlement Agreement. Even though the contract admits no ambiguity, the lower courts engrafted limitations that are entirely external to the instrument *770and directed to fulfilling Kleberg County's unexpressed subjective intent. This is not a proper use of surrounding facts and circumstances. "[C]ourts cannot rewrite the parties' contract or add to or subtract from its language."78
If the parties intended to obligate URI to restore the water quality in Well I-11 to use for irrigation purposes (or higher) or guarantee a minimum number of wells would be made suitable, the parties could have articulated so. But rather than agreeing to a particular result, the parties adopted a process that would be nugatory under Kleberg County's construction of the contract.79 For example, in section 11.1(1)(ii), the Settlement Agreement does not identify specific wells subject to restoration and, instead, leaves the matter subject to determination by reference to baseline data. If the parties intended to require Well I-11 to be restored to irrigation use, there would be no need for the open-ended description of the wells encompassed by section 11.1(1)(ii) as "90% or more of the combined total of TCEQ production area baseline wells in KVD Production Area 1 and any other wells in the production patterns that URI sampled and for which baseline is available before mining begins." To make an analogy, one would not mandate "delivery to the green house on Pecan Street" by agreeing that delivery could be made at "a colored house on a street with a nut name."
Subsection 11.1(2) would likewise be superfluous if the parties contemplated that specific data would be used or a specific outcome was guaranteed. In that subsection, the parties agreed that URI would provide "a copy of the analytical data" URI relied on in certifying which wells "are deemed to have produced water suitable for either drinking, stock watering, or irrigation use before URI's mining in PAA 1." That would be unnecessary if the parties already agreed on which data to use.
The agreement, construed as a whole, thus expresses an intent that URI would follow a process to identify subject PAA 1 wells, using unspecified baseline "analytical data," to certify which wells were "deemed to have produced" suitable water "before URI's mining in PAA 1 began." If the 1987 baseline had indicated a higher use for Well I-11 than irrigation, or resulted in more wells with prior suitability, URI would have been contractually bound to restore the water quality accordingly. The fact that no wells ultimately triggered the restoration requirement in section 11.1(1)(ii) does not render the provision or the substance of the Settlement Agreement meaningless. Kleberg County bargained for a process in section 11.1(1)(ii), and it got just that.
While the agreement is silent regarding averaging the 1985 and 1987 data, the record reflects averaging was a more conservative approach than relying solely on one set of data. According to the evidence at trial, the TCEQ requires only a single sample to establish baseline quality. Despite the industry practice of using only one sample to establish baseline quality, the testifying experts agreed that using multiple data points is a better measure of baseline quality than a single sample.
*771Though they disagreed as to the optimal number of data points, the record reflects that only two such samples existed for Well I-11. Thus, rather than relying on the 1987 data alone, URI reasonably averaged the available data.
Focusing on the term "available" in subsection 11.1(1)(ii), Kleberg County argues the Settlement Agreement linguistically fences out the 1987 data. The 1987 data existed before PAA 1 mining began in 1988 and before URI resumed PAA 3 mining in 2007,80 but Kleberg County construes "available" as meaning publicly available or otherwise known by the County at the time the Settlement Agreement was executed. Neither of Kleberg County's proposed limitations is expressed in the agreement, and extrinsic evidence of the parties' knowledge and expectations cannot be considered to add to, alter, or otherwise change the unambiguous language in subsection 11.1(1)(ii).81
Regardless, Kleberg County's reliance on its interpretation of "available" is misplaced. Well I-11 is indisputably subject to a restoration determination under subsection 11.1(1)(ii) because it is either a (1) "TCEQ production area baseline well[ ] in KVD Production Area 1" or (2) "any other well[ ] in the production patterns that URI sampled and for which baseline is available before mining begins [in PAA 3]." As a practical matter, it doesn't matter which one. What matters is URI's restoration obligation. Well I-11 must be "returned to suitability ... for the same use to which it was suited before such mining" only if it had "water suitable ... before URI's mining in PAA 1 for use as either drinking water, livestock water, or irrigation water ...." "Available" is pertinent to defining the subject PAA 1 wells but says nothing about the restoration obligation. Subsection 11.1(1)(ii) requires restoration of wells for which there is available baseline data, but beyond that, does not limit the baseline data for purposes of a suitability determination.
Kleberg County alternatively argues URI breached the Settlement Agreement because URI's section 11.1 certification did not, itself, comply with the contract requirements. Kleberg County alleges noncompliance in the following respects: (1) the sworn certification states that the analytical data showed unsuitability for contractually specified uses, rather than whether any wells had been suitable for those uses; (2) the certification is self-contradictory because it uses a double negative ("None of these wells were not suitable for use."); and (3) no documentation for the certification was submitted at trial.
We find no merit in this argument which relies on unpleaded theories of contract breach, inverts the burden of proof, and contradicts unchallenged trial court findings. The sworn certification and its accompanying transmittal letter state that the analytical data URI relied on was attached, as required by the Settlement Agreement. To establish URI's noncompliance with the contract requirements, Kleberg County bore the burden of proving the circumstances were otherwise.
*772With respect to the asserted inaccuracies in the certification, the Settlement Agreement requires only "a sworn statement of a URI officer certifying to the Kleberg County Judge" that pore treatment and required well restoration was completed before mining recommenced in PAA 3. Understanding these terms according to their normal usage, the agreement required URI's officer to swear under oath a formal assurance that URI met the contractual requirements.82 The agreement thus requires only that the certification be honest. The trial court found URI's alleged "breach" was unintentional and "without deliberate intent" and that "there was no bad faith on the part of either party with regard to [the] language interpretation issues and the parties' respective positions."
We hold that, under a plain and grammatical reading of section 11.1, URI did not breach the Settlement Agreement when it resumed mining operations in PAA 3, and Kleberg County is not entitled to attorney's fees or a specific-performance remedy on its breach-of-contract claim.83
III. Conclusion
The court of appeals relied on extrinsic evidence of intent to add to, alter, and augment the Settlement Agreement's plain and unambiguous language. This was clear error. As a matter of law, URI did not breach section 11.1 of the Settlement Agreement when it recommenced mining in PAA 3. We therefore reverse the court of appeals' judgment and render judgment that Kleberg County take nothing on its breach-of-contract claim.
Justice Blacklock did not participate in the decision.