Petitioner Raymond Espinoza, a criminal defendant in Maricopa County, challenges the policy adopted by a group of Maricopa County Superior Court judges of summarily rejecting all plea agreements containing stipulated sentences. The court of appeals affirmed the policy. We granted review, and we have jurisdiction pursuant to Ariz. Const, art. 6, § 5(3), and rule 31.19, Arizona Rules of Criminal Procedure.
FACTS AND PROCEDURAL HISTORY
The criminal divisions of the Maricopa County Superior Court are divided into four groups designated as quadrants A through D. Quadrant B, consisting of 5 judges, pre*146sides over felony offenses committed in the justice court precincts of Gila Bend, Tolleson, West Phoenix, and Maryvale. On December 28, 1992, the quadrant B judges issued a memorandum detailing a new plea agreement policy that was scheduled to take effect on January 25, 1993. The policy stated that quadrant B judges would no longer accept any plea agreements containing stipulated sentences because sentencing “is a judicial function which should not be subjected to limitations which are imposed by the parties, but are not required by law.” Although quadrant B had already adopted the policy, the memorandum indicated that the judges would welcome feedback. After holding a meeting to consider such feedback, the quadrant B judges issued a revised version of their policy on January 15, 1993 (quadrant B policy). The relevant section of that policy reads as follows:
1. Plea agreements may stipulate to “probation,” or “department of corrections” [DOC] for felonies, or “county jail” for misdemeanors. Agreements may not stipulate to any term of years (other than lifetime probation in dangerous crimes against children) or to any non-mandatory terms and conditions of probation (including, but not limited to, jail time, fines or surcharges, or community service hours), or to sentences running concurrently or consecutively, except for DOC time followed by lifetime probation in dangerous crimes against children.
The only 2 exceptions to the quadrant B policy are as follows:
2. Exceptions will be made for legitimate cooperation agreements. If the state wishes to make stipulated sentencing concessions in exchange for information, testimony or cooperation from a defendant, that fact should be made known to the judge in an appropriate manner prior to the change of plea.
4. Stipulations in capital murder cases to life imprisonment are viewed by the judges as charging concessions and not true sentencing stipulations. Therefore, such stipulations are unaffected by the policy.
On June 2,1993, Espinoza was indicted on one count of offering to sell narcotic drugs and one count of misconduct involving weapons. At his arraignment, the case was assigned to respondent, quadrant B Judge Gregory H. Martin. On August 11, 1993, Espinoza appeared before Judge Martin in chambers to enter a plea of guilty to both counts pursuant to a plea agreement, which stipulated that the sentences would run concurrently with each other and with an unrelated probation revocation. Judge Martin summarily rejected Espinoza’s plea agreement because the stipulation to concurrent sentences violated the quadrant B policy. On August 31, 1993, Espinoza presented Judge Martin with the same plea agreement, this time in court and on the record, and the judge again rejected the agreement without giving it any individualized consideration because, as he noted, “[tjhe sentencing provisions are all contrary to the Quad B policy.” (Emphasis added.)
Following Judge Martin’s ruling, Espinoza filed a petition for special action. The court of appeals accepted jurisdiction, but denied relief, holding that the quadrant B policy was a proper exercise of judicial authority. Espinoza v. Martin, 180 Ariz. 608, 886 P.2d 1364 (App.1993). Espinoza then filed a petition for review. We granted review to consider the validity of the quadrant B policy because it presents an issue of statewide importance that this court has not previously decided.
QUESTIONS PRESENTED
I. Whether the quadrant B policy violates rule 17.4, Arizona Rules of Criminal Procedure, because it prevents the trial court from exercising its discretion when deciding whether to accept or reject a plea agreement containing a stipulated sentence.
II. Whether the quadrant B policy violates rule 36, Arizona Rules of Criminal Procedure, because it establishes an unapproved local rule that is inconsistent with the Arizona Rules of Criminal Procedure.
*147DISCUSSION
I. Violation of Rule 174
Rule 17.4, Arizona Rules of Criminal Procedure, governs plea negotiations and agreements. This court has stated that “[t]he rules [of criminal procedure] recognize that properly negotiated plea agreements ... are an essential part of the criminal process and can enhance judicial economy, protect the resources of the State, and serve the ends of justice for the defendant, the State and the victim.” State v. Superior Court, 125 Ariz. 575, 577, 611 P.2d 928, 980 (1980). This case turns on the meaning of rule 17.4(a), which reads as follows:
Plea Negotiations. The parties may negotiate concerning, and reach an agreement on, any aspect of the disposition of the case. The court shall not participate in any such negotiation.
The plain language of rule 17.4(a) gives the parties the right to negotiate and reach agreement on “any aspect of the disposition of the case.” (Emphasis added.) This means that “the State and the defendant may bargain both as to the plea of guilty and as to the sentence to be imposed.” Superior Court, 125 Ariz. at 577, 611 P.2d at 930.
Although rule 17.4(a) allows the parties to negotiate plea agreements, including sentences, rule 17.4 also grants trial courts considerable discretion in deciding whether to accept or reject such agreements. State v. De Nistor, 143 Ariz. 407, 411, 694 P.2d 237, 241 (1985). Rule 17.4(d) provides in part:
Acceptance of Plea. After making such determinations [of the accuracy of the agreement and the voluntariness and intelligence of the plea] and considering the victim’s view, if provided, the court shall either accept or reject the tendered negotiated plea.
Furthermore, even if a trial court accepts a plea agreement, it is not bound by negotiated provisions regarding the sentence or the terms of probation if a review of the presentence report reveals the inadequacy of those provisions. Rule 17.4(d) and Form XVIII, Arizona Rules of Criminal Procedure.
In order to ensure that agreements negotiated pursuant to rule 17.4(a) have some meaningful effect, we interpret rule 17.4 as guaranteeing the parties the right to present their negotiated agreement to a judge, to have the judge consider the merits of that agreement in light of the circumstances of the case, and to have the judge exercise his or her discretion with regard to the agreement. Instead of hampering judicial sentencing discretion, the current version of rule 17.4, taken as a whole, contemplates the exercise of judicial discretion when determining whether to accept or reject each particular plea agreement. In exercising that discretion, the trial court must “review the plea agreement to see if the ends of justice and the protection of the public are being served by such agreement.” Superior Court, 125 Ariz. at 577, 611 P.2d at 930.
Therefore, the court of appeals correctly noted that “[t]he language in rule 17.4(a), allowing the parties to ‘negotiate’ and ‘agree’ upon any aspect of the case, does not give the parties or the lawyers the right to force unwilling judges to accept plea agreements with sentence stipulations.” Espinoza, 180 Ariz. at 613, 886 P.2d at 1369. However, the court of appeals erroneously believed that allowing the parties to negotiate stipulated sentences, pursuant to rule 17.4(a), conferred a “binding sentencing power” on the prosecutor in violation of the separation of powers. 180 Ariz. at 614, 886 P.2d at 1370. After giving full consideration to the appropriateness of a plea agreement, the trial court has the discretion to either accept or reject the entire plea agreement, or to accept the agreement and later reject the sentencing provisions if deemed inappropriate after further inquiry. Therefore, there was no need for the court of appeals to try to further enhance judicial sentencing discretion by approving a policy that limited the parties’ right to negotiate.
This court has previously invalidated policies that limit the exercise of judicial discretion in accepting plea agreements. In the analogous case of Hare v. Superior Court, this court struck down a guideline in part because it conflicted with rule 17.4. 133 Ariz. 540, 542-43, 652 P.2d 1387, 1389-90 (1982). Guideline B, adopted by the Pima County Superior Court, provided that “after *148the first trial date, no plea agreements would be accepted by the court except those which result in pleas to the charges contained in the indictment.” Hare, 133 Ariz. at 541, 652 P.2d at 1388. In holding that guideline B exceeded the limits imposed on the superior court by rule 17.4(a), we reasoned that rule 17.4
recognizes not only the right to engage in plea negotiation and agreement, but provides for the procedure by which the judge rejects or accepts a plea agreement. Rules 17.4(d) & (e) allow the judge to exercise his discretion when a plea agreement is presented for approval. Under Guideline B, the court is prohibited from exercising that discretion----
Hare, 133 Ariz. at 542, 652 P.2d at 1389. In Hare, guideline B prevented the trial court from considering plea agreements after a certain date. In this case, quadrant B policy prevented the trial court from considering plea agreements with stipulated sentences. The end result is the same. The quadrant B policy, like guideline B in Hare, violates rule 17.4 because it precludes a judge from exercising his discretion over plea agreements in a predetermined set of circumstances.
Espinoza, his attorney, and the prosecutor negotiated pursuant to rule 17.4(a) and reached a plea agreement. Espinoza agreed to plead guilty to two charges: attempting to knowingly sell a narcotic drug and knowingly possessing a deadly weapon during the commission of a felony. In exchange, the parties agreed that the sentences imposed on both charges would be served concurrently, and that those sentences would also be concurrent with a probation revocation.
When Espinoza presented Judge Martin with this plea agreement on two different occasions, the judge summarily rejected it both times because the stipulation to concurrent sentences violated the quadrant B policy. Judge Martin did not consider the particular circumstances of the case and made no findings regarding the appropriateness of the negotiated sentence. Instead, the presence of a stipulated sentence in the agreement triggered the quadrant B policy and precluded any individualized exercise of discretion. Absent the quadrant B policy, Judge Martin could have weighed the merits of the plea agreement and accepted it, rejected it entirely, or rejected the sentencing provisions as inappropriate once he had reviewed the presentence report. This is the type of discretion contemplated by rule 17.4, and trial courts are obligated to exercise it.
Because the quadrant B policy simultaneously limits the content of plea agreements and precludes the exercise of judicial discretion over individual plea agreements, we hold that the policy violates rule 17.4. While courts are free, pursuant to rule 17.4(d), to reject plea agreements with stipulated sentences after giving them individualized consideration, groups of judges may not implement policies to automatically reject all such plea agreements without considering whether a stipulated sentence is appropriate in light of the circumstances of the case.
Our holding applies equally to the actions of individual judges. The respondent relies on State ex rel. Bowers v. Superior Court in which the court of appeals held that a trial judge in Navajo County, acting alone and not pursuant to a written policy, did not abuse his discretion by summarily rejecting a plea agreement because it contained a stipulated sentence. 173 Ariz. 34, 40, 839 P.2d 454, 460 (App.1992). We denied review in Bowers, and we now express our disapproval of that case to the extent that it allows a trial judge to automatically reject a plea agreement without individualized consideration because it contains a stipulated sentence.
II. Violation of Rule 36
As noted above, we hold that the quadrant B policy violates rule 17.4, Arizona Rules of Criminal Procedure. We further hold that the quadrant B policy violates rule 36, Arizona Rules of Criminal Procedure, because quadrant B adopted a rule that is inconsistent with the Arizona Rules of Criminal Procedure. Even if the quadrant B policy were consistent with the rules of procedure, the policy constituted a local rule that was invalid because quadrant B adopted it without first obtaining the approval of this court.
This court has the exclusive power to make rules pertaining to all procedural mat*149ters in any Arizona court. Ariz. Const, art. 6, § 5(5). This rulemaking power “may not be supplemented, annulled or superseded by an inferior court.” Anderson v. Pickrell, 115 Ariz. 589, 590, 566 P.2d 1335, 1336 (1977). However, rule 36 provides that:
Any court may make and amend rules governing its practice not inconsistent with these rules. No such rule shall become effective until approved in writing by the Supreme Court.
This issue turns on the meaning of the word “rule” as used in rule 36, because quadrant B was required to obtain this court’s approval if the policy it adopted constituted a procedural rule. The court of appeals held that the quadrant B policy did “not rise to the level of a procedural rule” and therefore did not require the approval of this court. Espinoza, 180 Ariz. at 616, 886 P.2d at 1372.
“A rule of court prescribes a procedural course of conduct that litigants are required to follow, the failure to comply with which may deprive the parties of substantial rights.” Hare, 133 Ariz. at 542, 652 P.2d at 1389. Hare involved a set of guidelines adopted by the Pima County Superior Court to facilitate its Automated Calendaring Project. In that case, we held that a guideline requiring judges to automatically reject plea agreements in certain circumstances was invalid in part because it was an unapproved local rule. Hare, 133 Ariz. at 542, 652 P.2d at 1389. The court reasoned that the Pima County guidelines
prescribe a course of conduct for certain aspects of the criminal practice in Pima County Superior Court. They are, in effect, local rules of criminal procedure which parties to criminal actions in Pima County must follow or lose substantial rights, in this case the right to have a plea bargain considered by the court.
Hare, 133 Ariz. at 542, 652 P.2d at 1389. This court relied on Hare in a later case when we held that an order by the chief city magistrate of Tucson requiring the prosecutor to make an avowal of good faith before filing for a change of judge was a local rule that was invalid for lack of this court’s approval. State v. City Court, 150 Ariz. 99, 103, 722 P.2d 267, 271 (1986).
Under the standard set forth in Hare, the quadrant B policy is a local rule of procedure. Just as in Hare, the policy adopted by quadrant B requires the participating judges to automatically reject plea agreements under certain circumstances. The quadrant B policy requires litigants to omit all stipulated sentences from their plea agreements, and it deprives the parties of the right to have the judge consider their plea agreements on a case-by-case basis.
In Hedlund v. Sheldon, we held that a trial judge’s decision to impanel dual juries for co-defendants was not a rule of procedure, but rather was “ ‘the exercise of an individual judge’s discretion to use a particular technique in order to meet a specific problem’ in a single case.” 173 Ariz. 143, 146, 840 P.2d 1008, 1011 (1992), quoting State v. Lambright, 138 Ariz. 63, 78, 673 P.2d 1, 16 (1983) (Feldman, J., specially concurring), cert. denied, 469 U.S. 892, 105 S.Ct. 267, 83 L.Ed.2d 203 (1984). Here, the respondent judge cites Hedlund as support for his assertion that the quadrant B policy is not a rule, but instead involves a group of judges jointly deciding to use a discretionary technique.
First, Hedlund is applicable only when a trial judge adopts a technique that is consistent with the rules of this court. The quadrant B policy violates rule 17.4 and thus is inconsistent with the rules of this court. Moreover, Hedlund is easily distinguishable from this case because it involved one judge using a particular technique to solve a problem in a single case. The quadrant B policy, on the other hand, is an agreement among a group of judges to apply a particular procedure, under» a predetermined set of circumstances, in all cases that come before those judges. Hedlund does not change our opinion that, under Hare, the quadrant B policy is a local rule within the meaning of rule 36.
Having concluded that the quadrant B policy is a rule, we hold that the policy is therefore invalid because the quadrant B judges failed to obtain the approval of this court before adopting it. If we allow quadrant B, or any other faction of superior court judges, to adopt its own rules without first coming to this court for approval, proce*150dure within the Arizona court system would become balkanized. Under Bowers and the quadrant B policy, one judge in Navajo County and 5 judges in Maricopa County refused to accept any plea agreement containing a stipulated sentence and refused to consider the agreement and the stipulated sentence on a case-by-case basis. In all other courts, judges presumably give individualized consideration to all plea agreements, even those with stipulated sentences. The purpose of rule 36 is to promote procedural uniformity, and unapproved local rules that conflict with statewide rules of procedure contravene that purpose.
At oral argument, respondent judge alleged that the quadrant B policy was simply an experiment, which is common practice in superior courts. However, no evidence in the record supports this theory. Once the quadrant B policy took effect, no one gathered statistics or conducted contemporaneous interviews' with interested parties on an ongoing basis to track the results of the “experiment.” In addition, no control groups were identified for comparison and no time limit was set for the project. The judges participating in the quadrant B “experiment” were not randomly selected, but instead they self selected into the project presumably because they favored the quadrant B policy. In short, the quadrant B judges cannot now claim that their policy was an experiment when they failed to follow any accepted experimental methodology. See generally Experimentation in the Law: Report of the Federal Judicial Center Advisory Committee on Experimentation in the Law 15-23, 71-76, 81-121 (Fed.Judieial Ctr.1981); David P. Farrington, Randomized Experiments on Crime and Justice, in 4 Crime and Justice: An Annual Review of Research 257, 271-76, 296-98 (Michael Tonry & Norval Morris eds., 1983); John Monahan & Laurens Walker, Social Science in Law: Cases and Materials 57-61 (1994).
Even if the quadrant B policy were a legitimate experiment, the judges in quadrant B are still subject to the provisions of rule 36. In Hare, we held that the guidelines adopted in Pima County violated rule 36, even though they were part of a superior court experiment to promote trial certainty and reduce delay. 133 Ariz. at 541, 543, 652 P.2d at 1388, 1390. Although we find that the adoption of the quadrant B policy is not a legitimate experiment, “[n]othing we say here should discourage courts, through the adoption of local rules, to carry out experiments which may improve the judicial process. Indeed, these efforts should be encouraged. But local rules must first be approved by this court before they are effective.” Hare, 133 Ariz. at 543, 652 P.2d at 1390.
Defendant makes a final claim that the quadrant B policy violates the equal protection clause of the Fourteenth Amendment of the United States Constitution because it treats similarly situated defendants differently depending on the geographic location of the crime scene. Furthermore, the policy subjects defendants in quadrant B to procedures different from those made applicable to other quadrants by rule 17.4. Because we conclude that the quadrant B policy conflicts with the Arizona Rules of Criminal Procedure, it is unnecessary to decide this constitutional issue. State v. Yslas, 139 Ariz. 60, 63, 676 P.2d 1118, 1121 (1984).
III. Justice Martone’s Dissent
The dissent warrants a brief mention. Justice Martone argues that the quadrant B policy is valid because “[i]f, after looking at the document, a judge is opposed to any part of the plea, he or she may summarily reject it [pursuant to rule 17.4], And that is precisely what Judge Martin did here.” Espinoza v. Martin, 182 Ariz. 145, 146, 894 P.2d 688, 689 (1995) (Martone, J., dissenting). However, the quadrant B policy required Judge Martin to reject Espinoza’s plea agreement whether or not he actually looked at the document. The dissent fails to explain how a policy that categorically predetermines the fate of certain plea agreements is consistent with rule 17.4, which requires an individualized review of each and every plea agreement.
The dissent further argues that, according to rule 17.4, Judge Martin “was free to accept or reject [the quadrant B policy], altogether or in a specific case.” Espinoza, 182 Ariz. at 148, 894 P.2d at 691. Whether *151Judge Martin had the discretion to accept or reject the quadrant B policy is irrelevant to an analysis of rule 17.4, which deals with the exercise of discretion over individual plea agreements and not over judicial policies. Furthermore, the dissent refuses to acknowledge that once Judge Martin accepted the quadrant B policy as a whole, he was not free to reject the policy “in a specific case.” The whole point of the quadrant B policy was to lock the participating judges into a predetermined course of action in the large majority of cases, such as this one, where the narrow exceptions to the quadrant B policy do not apply. That is why the policy precludes the exercise of discretion over individual plea agreements that rule 17.4 contemplates. If, as Justice Martone suggests, Judge Martin were merely following rule 17.4, then there was no need to adopt the quadrant B policy.
In light of our holding in this case, the dissent asks “what of the Rule V Inactive Calendar Guidelines and the Guidelines for Rule 26.1” adopted by the Maricopa County Superior Court. Espinoza, 182 Ariz. at 154 n. 1, 894 P.2d at 697 n. 1 (citations omitted). The obvious distinction is that both sets of guidelines, by their terms, were adopted in an attempt “to facilitate uniform and predictable application” of rules 26.1 and V—not to negate those rules.
Although the dissenting opinion refers to the quadrant B policy as “the quadrant B experiment,” this label is meaningless because the dissent fails to offer any facts proving up the legitimacy of the so-called experiment. Espinoza, 182 Ariz. at 154 n. 1, 894 P.2d at 697 n. 1. When a court of this state wants to conduct a legitimate experiment, which is at variance with existing rules of court, that court should present a proposal to this court for approval before the experiment commences.
DISPOSITION
We hold that the quadrant B policy of summarily rejecting plea agreements with stipulated sentences violates both rules 17.4' and 36, Arizona Rules of Criminal Procedure. Therefore, we find that the court of appeals erred when it approved this policy and denied the relief requested in defendant’s petition for special action. We reverse the trial court, vacate the court of appeals’ decision, and remand this case to the trial court so that it may fully review the appropriateness of the plea agreement presented by the county attorney and the defendant and exercise its discretion in accepting or rejecting that agreement.
MOELLER, V.C.J., concurs.