Opinion for the court filed by Circuit Judge REYNA.
Dissenting opinion filed by Circuit Judge MOORE.
REYNA, Circuit Judge.Nissim Corp. (“Nissim”) seeks to enforce an agreement it entered into to settle its patent infringement lawsuit against ClearPlay Inc., Matthew Jarman, and Lee Jarman (collectively, “ClearPlay”). This is at least the third time the instant case has come before this court.1 Because the district court’s dismissal of this case as moot was error, we reverse and remand for further proceedings.
BACKGROUND
I.
Nissim holds patents enabling viewers of DVDs to skip or mute objectionable content (sometimes abbreviated “OC”), and has developed the CustomPlay Specifications (“Specifications”) to standardize the creation of “filters” for that purpose. These filters allow viewers with compatible players to skip or mute content that they find objectionable. To create a filter for a DVD, a provider must associate scenes in the movie with codes from the Specifications. For example, a filter could indicate that time range 0:32:00 — 0:32:15 of the DVD corresponds to the Specifications’ code for explicit violence. This process is called “coding.”
ClearPlay developed a DVD player capable of filtering objectionable content, as well as filters for that player, and Nissim sued ClearPlay for infringement. The parties entered into an agreement pursuant to which ClearPlay would obtain a license in exchange for agreeing to implement its filters in compliance -with the Specifications, and the case was dismissed. At the time of settlement, ClearPlay had produced over 2000 filters.2
Nissim eventually came to believe that ClearPlay had not brought the filters into compliance. It filed a motion to enforce the agreement, and the district court appointed a special master to determine whether ClearPlay’s filters complied with the Specifications. Rather than examine thousands of filters, the parties agreed to a procedure by which the special master would examine a representative sample of eight filters. Four of the eight sample filters were selected by Nissim. The remaining four were chosen by the special master from a list of ten filters chosen by ClearPlay. After viewing the movies and hearing testimony, the special master recommended denial of the motion to enforce, finding that the filters in the sample sub*25stantially complied with the requirements of the agreement. The district court adopted the recommendation.
Nissim appealed, and we vacated the district court’s judgment. This court was concerned that interpreting the “artistic judgment” portion of the agreement expansively, as the district court did, would swallow the substantial compliance rule. It vacated the order and remanded “for the fact finder to determine whether ClearPlay’s OC maps substantially comply with the Specifications.” Nissim I, 374 Fed.Appx. at 994.
II.
On remand, Nissim again filed for summary enforcement of the agreement, requesting that the court “enjoin ClearPlay from selling or distributing the filters it created under the wrong standard.” Nis-sim contended that ClearPlay’s systemic failure to use the Specification in crafting its filters meant that ClearPlay was not in substantial compliance under the terms of the agreement. The court denied the motion, observing that this court’s remand opinion required determinations of filter compliance to be performed on a case-by-case basis. The court again referred the matter to the special master, noting that it would “be more efficient and fair to use the same procedure as before, with the results from a representative sample set of filters being presumed to apply to all the filters from which the sample was obtained.” J.A. 4.
ClearPlay moved to resume proceedings before the special master, and Nis-sim argued that a representative sample procedure was no longer appropriate. According to Nissim, this court’s opinion in Nissim I precluded the court from using such a procedure. Nissim therefore suggested that the special master’s recommendation should be “limited to the individual filters themselves and not presumptively applied as a whole to the universe of filters from which the sample set was obtained.” J.A. 5899. Ultimately, the court ordered the proceedings to be resumed before the special master, directing him to examine the same eight filters he had previously examined and to produce a report oh “whether each of the [filters, separately and individually, substantially complies with the CustomPlay Objectionable Content Specifications.” J.A. 5930. While the revised order governing the special master proceedings no longer refers to a “representative sample,” it also contains nothing to contradict its post-remand statement that it would again employ a representative sample methodology.
ClearPlay then moved to dismiss the action as moot, representing that it had discontinued sales of the eight filters in the sample set and arguing that there was nothing further for the court to decide in this case. Nissim replied, making the same arguments it makes to this court. Notably, Nissim argued that the revised special master procedure “did not purport to alter the sample-set approach that had been previously ordered.” J.A. 5999.
The court heard argument on the mootness motion. It began by stating its belief that the case was moot because “if the case is dismissed [ClearPlay] will either comply with Nissim’s request that it reconfigure the filters in question, or just discard them, and not sell them.” J.A. 9286. The “filters in question” appears to refer to the eight filters examined by the special master that ClearPlay voluntarily stopped selling and distributing. The court explained its understanding that our remand opinion in Nissim I precluded it from deciding the case using a representative sampling, and that the correct way to proceed was via the mediation procedures in section 4.6 of *26the settlement agreement. The court then stated that it had “already indicated that ... other than the [filters] that were the subject of the submission to the special master [it] was not going to deal with any other filtering issues.” J.A. 9288. It concluded that, since ClearPlay had offered to implement Nissim’s changes to those eight filters or withdraw them from distribution, there was not “anything else for this Court to do.” J.A. 9288. The court emphasized its desire to be done with the case, stating: “Any future filter disputes, do what you want, I am not going to handle them. I would assume you would handle it pursuant to section 4.6 of the settlement agreement, which is how it should have been done in the first place.”3 J.A. 9295-96. On the same day, the court entered its order dismissing the case as moot “for the reasons discussed in open court.” J.A. 6219.
Nissim appeals. We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(1).
DISCUSSION
Nissim raises two arguments: first, that the district court erred in denying its “Motion for Summary Enforcement” of the Agreement, and second, that the district court erred in dismissing the case as moot. Because ClearPlay contends that mootness deprives this court of jurisdiction over the appeal, we address that issue first.
I.
Nissim argues that it was error for the district court to dismiss the case as moot. The thrust of Nissim’s argument is that although ClearPlay withdrew eight of the filters, there was still a case or controversy regarding the remaining 1992 filters. Nis-sim also points out that ClearPlay’s withdrawal of eight filters falls far short of the relief that Nissim seeks. ClearPlay responds that the parties had agreed to limit this ease to just the eight filters and that the district court correctly concluded that ClearPlay’s decision to discontinue those filters rendered the case moot.
The concept of “mootness” derives from the case or controversy requirement of Article III of the Constitution. See Allen v. Wright, 468 U.S. 787, 750, 104 S.Ct. 3315, 82 L.Ed.2d 556 (1984). “Mootness is a jurisdictional question because the Court is not empowered to decide moot questions or abstract propositions^]’ ” North Carolina v. Rice, 404 U.S. 244, 246, 92 S.Ct. 402, 30 L.Ed.2d 413 (1971) (internal quotation marks omitted). To avoid mootness, “[t]he controversy must be definite and concrete, touching the legal relations of parties having adverse legal interests.” Aetna Life Ins. Co. of Hartford, Conn. v. Haworth, 300 U.S. 227, 240-41, 57 S.Ct. 461, 81 L.Ed. 617 (1937). The case must be real and substantial, “admitting of specific relief through a decree of a conclusive character, as distinguished from an opinion advising what the law would be upon a hypothetical state of facts.” Id. at 241, 57 S.Ct. 461. “[T]o avoid dismissal for mootness, an actual controversy must remain at all stages, not merely at the time the complaint is filed.” Gerdau Ameristeel Corp. v. United States, 519 F.3d 1336, 1340 (Fed.Cir.2008) (citing Steffel v. Thompson, 415 U.S. 452, 460 n. 10, 94 S.Ct. 1209, 39 L.Ed.2d 505 (1974)). “Whether an actual controversy exists to support subject matter jurisdiction is a question of law subject to de novo review.” Highway Equip. Co., Inc. v. FECO, Ltd., 469 F.3d 1027, 1031-32 (Fed.Cir.2006).
The district court dismissed this case as moot “for the reasons discussed in open *27court.” J.A. 6219. But a review of the transcript of the hearing reveals little more than that the court was exasperated with the parties. Having experienced the parties and attorneys in this case, we are not unsympathetic.4 Nevertheless, whether the case is moot does not depend on a court’s opinion of the litigants or of counsel.
In light of our review of the transcript of the hearing, we understand the legal basis of the court’s decision to be its belief that ClearPlay’s concessions regarding the eight filters left it with no remaining issues to resolve. This belief was incorrect for at least two reasons. First, Nissim sought more than just an order requiring ClearPlay to use the specification in making its filters. In its original motion, Nissim sought to enjoin the sale of ClearPlay’s players and to enjoin the sale and distribution of ClearPlay’s filters. On remand, Nissim again asked the court to enjoin ClearPlay from selling or distributing the filters. It referred to this relief in its response to the motion to dismiss. Although Nissim did not identify this relief in its briefing, it did reiterate this request at oral argument. Oral Arg. at 17:30, avaHable at http://oralarguments.cafc. uscourts.gov/default.aspx?fl=2012-1188. mp3.
Second, Nissim sought relief on all of the filters, not just the eight in the sample. The representative sampling that was originally performed examined eight filters in order to determine compliance for all of the pre-settlement filters. After Nissim I, the district court remained committed to using sampling rather than examining all 2000 filters. Although it did not contain language about a “representative sample,” the updated special master order pertained only to the eight movies originally sampled. The court’s decision to submit only those eight filters to the special master did not divest it of its jurisdiction over the remaining filters. Nor is it inconsistent with the court’s prior statement that it would continue to employ a representative sampling technique. In our view, the updated order reflects the fact that the sample had already been chosen and that the district court was taking care to conduct the special master investigation in a manner consistent with our opinion.
To the extent that the court may have interpreted our opinion in Nissim I to *28preclude it from an employing a representative sampling to determine compliance for the entire set of filters, that was also error. Although Nissim I emphasized the need for the fact finder to address each movie on a case-by-case basis, it does not follow that the fact finder was required to examine each of the 2000 movies individually. Nissim I held only that the “artistic judgment” language of section 1.4 did not create a blanket exception to compliance with the Specifications. 374 Fed.Appx. at 993 (holding that section 1.4 does not “allow ClearPlay to use its artistic judgment to refuse to code objectionable material because it believes the relevance to outweigh the objectionableness.”). Rather, the decisions to code or not to. code particular content must be based on the fact-specific nature of the content at issue. This does not mean that the court and the parties may not agree to use a representative sampling methodology to settle the dispute without having to evaluate all 2000 movies — they may, so long as the evaluation of the representative filters includes the fact-specific inquiries required by the agreement and the Specifications. We would also expect that any sampling methodology used is one that is agreed upon by the parties and the court concerning the selection and number of filters that will be reviewed. That appears to be what happened in this case. Indeed, the court and the parties agreed that the results of the review of the eight sample filters would be made applicable to the entire universe of filters, i.e., the 2000, and the court reiterated that procedure on remand from Nissim I.
Our review of the record also reveals no adequate basis to conclude that the court had withdrawn jurisdiction over the remaining 1992 filters. The court’s original order withdrawing jurisdiction was vacated by this court. Nissim I, 374 Fed.Appx. at 994. After remand, the court’s order denying Nissim’s motion to enforce did not withdraw jurisdiction, and in fact continued to apply a representative sampling procedure. The court also denied Nissim’s motion to reconsider, citing “the reasons stated in open court at the November 23rd hearing.” J.A. 5812. The transcript of that hearing consumes over fifty pages and contains two references to jurisdiction, each of which states that the court had indicated in a prior order that it had terminated its jurisdiction to resolve any further filter disputes. But the only prior order to that effect in the record is .the order that was vacated by this court. And this court had already said that the court’s attempt to relinquish jurisdiction over future disputes would not be ripe until “the parties filed a second dispute over the Agreement with the district court and the district court refused to assert jurisdiction.” Nissim I, 374 Fed.Appx. at 994. Given the extensive discussion in that hearing, we do not believe the parties would be on notice that the motion to reconsider had been denied based on those two passing references to a vacated order.
Subsequently, ClearPlay took the position that Nissim’s disputes as to the remaining' 1992 were “future” disputes. On that basis, it discontinued the eight sample filters and filed a motion to dismiss the case as moot. The court held a hearing on ClearPlay’s motion in which jurisdiction was alluded to twice. First, apparently rejecting Nissim’s argument that Clear-Play had waived arbitration, the district court stated that filter disputes should be settled pursuant to section 4.6, and that “other than the [filters] that were the subject of the submission to the special master[, it] was not going to deal with any other filtering issues.” J.A. 9288. Second, based on its incorrect understanding that Nissim I precluded the use of a representative sample, and in contravention to Nis-*29sim I’s mandate requiring “the fact finder to determine whether ClearPlay’s OC maps substantially comply with the Specifications,” 374 Fed.Appx. at 994, the court stated that it was “no longer going to retain jurisdiction to enforce the settlement agreement.” J.A. 9290.
The district court might well have had discretion to dismiss the case entirely, but that is not what it did. Instead, the court reframed the dispute from a representative sample to a dispute over eight individual filters and characterized the dispute as to the remaining 1992 filters as a future dispute. We recognize that the need to interpret the orders in light of the extensive “proceedings in open court” leaves room for debate as to the actual basis for the court’s decisions;5 however, on the record before us, we decline to read the transcript as an exercise of the district court’s discretion to withdraw jurisdiction over the existing controversy over the 2000 filters. Nissim’s requests for relief with regard to the remaining 1992 filters was not a “future dispute.” It was before the court and could easily have been resolved by the representative sample had Clear-Play not frustrated that effort by withdrawing the sample filters, throwing out years of work. But the controversy still remained, and the decision to dismiss this case on the basis of mootness was error.
II.
Having concluded that the case is not moot, we turn to Nissim’s argument that the district court erred in denying its motion for summary enforcement. Nissim argues, as it did below in its motion for summary enforcement, that the undisputed evidence shows that ClearPlay did not base its coding decisions on the Specifications. The district court denied Nissim’s motion, observing that “the test for whether ClearPlay’s filters are compliant is whether ClearPlay gets the right result, not how it gets there.” J.A. 5388. Citing this court’s instruction that compliance must be determined on a case-by-case basis, the district court determined that it could not enjoin ClearPlay’s sales and distributions of all filters based on testimony about the process by which the filters had been produced.
In light of our holding that this case is not moot, we conclude that we lack jurisdiction to address this issue. Our jurisdiction is limited by statute to appeals from final decisions. See 28 U.S.C. § 1295(a)(1) (providing this court with jurisdiction “from a final decision of a district court” in cases arising under the patent laws). To be appealable, a decision must be “a final dispositive ruling that ends litigation on the merits.” Copelands’ Enters., Inc. v. CNV, Inc., 887 F.2d 1065, 1067 (Fed.Cir.1989) (en banc) (quoting R.G. Barry Corp. v. Mushroom Makers, Inc., 609 F.2d 1002, 1005 (CCPA 1979)). The order denying Nissim’s motion for summary enforcement did not end the litigation on the merits and therefore was not final. See, e.g., Lermer Germany GmbH v. Lermer Corp., 94 F.3d 1575, 1576-77 (Fed.Cir.1996) (holding that the denial of summary judgment is not a final decision subject to appeal); Aleut Tribe v. United States, 702 F.2d 1015, 1019 (Fed.Cir.1983) (holding that the denial of a motion to dismiss is not a final decision subject to appeal). The court has yet to resolve Nis-sim’s complaint regarding the remaining 1992 filters, as well as the remaining remedies Nissim requests. Only when it has done so will we have a decision to review.
*30III.
In the interests of clarity, it is important to identify several issues which are not properly before us on appeal and which we therefore do not decide, but which may be relevant on remand.
We do not decide whether section 4.6 of the agreement requires the parties to arbitrate their filter disputes. Although the court seems to have discussed arbitration on numerous occasions, the parties never requested a decision on this issue and none was granted. Similarly, we do not address Nissim’s argument that arbitration was waived, as the district court has not decided this issue one way or the other. Nor did the court’s order refer the remaining filters to arbitration under section 4.6 of the agreement, as ClearPlay contends. Nothing prevents either party from moving to compel arbitration on remand; however, the issue is not before us at this time.
We also express no opinion as to the merits of the remaining 1992 filters. All that can be said at this point is that the eight filters withdrawn by ClearPlay are no longer in the case. The district court has made no decision as to whether any of the remaining filters comply with the Specifications. Nor has it determined by what means this dispute as to those filters will be resolved. As discussed above, our opinion in Nissim I does not preclude the court from using a representative sample. The district court might, for example, determine that ClearPlay’s withdrawal of the eight samples operated as a concession that all of the filters were noncompliant. On the other hand, it might select another representative sample of filters. We decline to speculate as to what route the court will choose; these issues all remain to be addressed, if at all, on remand.
Finally, because we read the court’s order as relinquishing jurisdiction only over future disputes — not over the current dispute in the instant case — we do not decide under what circumstances it would be an abuse of discretion for the district court to limit or relinquish the jurisdiction it retained over the settlement agreement in its original order dismissing the patent case, and which it has exercised for the past six years. See Kokkonen v. Guardian Life Ins. Co. of Am., 511 U.S. 375, 380-81, 114 S.Ct. 1673, 128 L.Ed.2d 391 (1994) (holding that a federal district court has the discretion to retain ancillary jurisdiction to enforce a settlement agreement).
IV.
We reverse the district court’s dismissal of this case for mootness, and remand for further proceedings in accordance with this opinion.
REVERSED
Costs
Each party shall bear its own costs.
. See In re Nissim Corp., 428 Fed.Appx. 981 (Fed.Cir.2011) (denying Nissim’s request for mandamus); Nissim Corp. v. Clearplay, Inc. ("Nissim I"), 374 Fed.Appx. 987 (Fed.Cir.2010) (interpreting the settlement agreement). We have also previously declined to hear ClearPlay’s appeal of its related case against Nissim for tortious interference with a contractual relationship. ClearPlay, Inc. v. Abecassis, 602 F.3d 1364 (Fed.Cir.2010).
. It is not clear from the record, nor does it matter to this appeal, exactly how many filters are at issue. The parties have used the number 2000 as an approximation, and we shall do the same.
. ClearPlay did make an oral motion to enforce the agreement through section 4.6, which the district court denied. See J.A. 9087-88.
. The excessive hyperbole in the briefs makes them difficult to take seriously and unpleasant to read, and strips both parties of their credibility. Nissim characterizes ClearPlay's arguments as "moanfing],” "excuse[s](" and "absurd”; it describes one of ClearPlay's communications with the special master as "bias-inducing screed.” ClearPlay’s briefs are no better; disparaging Nissim’s suit as "unnecessarily time-consuming and expensive,” "ill-conceived,” "wasting] the time and resources of ClearPlay and the [c]ourt,” and a "massive waste of judicial time and resources,” and referring to Nissim’s arguments "inexplicable,” "strange,” and "baffling." The record reveals the parties’ behavior in the district court proceedings to be even worse. The parties would be well-advised to take the advice of Justice Scalia and Bryan Gamer: “Cultivate a tone of civility, showing that you are not blinded by passion.... A straightforward recital of the facts will arouse whatever animosity the appellate court is capable of entertaining, without detracting from the appearance of calm and equanimity that you want to project.” Anto-nin Scalia & Bryan A. Garner, Making Your Case: The Art of Persuading Judges 34-35 (2008).
The parties' poor judgment extends beyond their choice of words. For example, Nissim stated that "the district court stuck Nissim with ClearPlay’s trivial gesture” and that the court’s decision ”mock[ed] justice.” These statements were at best unprofessional. See ABA Model Rules of Professional Conduct Rules 3.5, 8.4(d). ClearPlay’s statements throughout its brief that Nissim had requested only prospective relief were obvious misrepresentations that cannot easily be explained away as mistakes. See id. Rule 3.3(a).
. Should the district court again choose to dismiss this case on remand, regardless of the basis, we encourage the court to set forth its reasoning in a written order.