Vacated in part, reversed in part, and remanded in part by published opinion. Judge KING wrote the opinion, in which Chief Judge WILKINSON and Senior Judge BUTZNER joined.
OPINION
Patricia J. Mayes appeals from the district court’s dismissal of her case pursuant to Federal Rule of Civil Procedure 12(b)(6) for failure to state a claim upon which relief can be granted. Mayes initially filed suit in Maryland state court, but the original defendants, Stanley and Judith Rapoport (collectively, “the Rapo-ports”), removed the case to the District of Maryland. Mayes subsequently filed an amended complaint naming David Key, a nondiverse party, as a defendant. The district court dismissed Key from the case, relying upon decisions applying the doctrine of fraudulent joinder, then dismissed the case against the Rapoports.
Mayes claims that the district court erred both in holding that it possessed jurisdiction and in dismissing the Rapo-ports under Rule 12(b)(6). We agree with Mayes that the Rapoports did not establish that Mayes fraudulently joined Key as a defendant and that the district court erred in its dismissal of Key. We therefore vacate the district court's judgment, reverse its holding that it possessed jurisdiction, and remand for further remand to state court.
I.
A.
Mayes filed this suit in Maryland state court to enforce her “right of first refusal to match an offer to purchase” commercial *459property at 1720 and 1722 Thames Street in Baltimore, Maryland (“property”). Mayes had been leasing that property from the Rapoports, and a provision of the lease had guaranteed her a right-of-first-refusal. The lease also provided that, “Should the tenant exercise her option to purchase said property she will be credited with $400 per month of occupancy toward the purchase price. This $400 rebate will be limited to only the first year of tenancy.”
In April 1997, the Rapoports received an offer to purchase the property from Key. Key submitted a proposed contract to the Rapoports that offered $400,000 for the property, with an $80,000 down payment, and the $320,000 balance financed by the Rapoports at 9% interest. Key’s offer also provided for a “balloon payment” at the end of 6 years and stipulated that Key would deposit $5,000 at the time of the offer and an additional $10,000 within 5 days of acceptance of the contract.
On June 13, 1997, before presenting it to Mayes under the right-of-first-refusal, the Rapoports accepted Key’s offer. However, the Rapoports’ signed contract with Key (the “Key contract”) did not explicitly condition the Rapoports’ acceptance upon Mayes’s right-of-first-refusal.1
Fifteen days later, on June 28, 1997, Mayes received a copy of the Key contract from the Rapoports. On July 1, 1997, Mayes finished her first year of tenancy, thus completing her deposit of $4,800 toward purchase of the property. She then carried-over as a tenant, which meant, under the lease, that Mayes was leasing the property from the Rapoports month-to-month under the same terms and conditions as in the original lease, including the right-of-first-refusal. Also on July 1, 1997, Mayes submitted an offer to the Rapoports attempting to exercise her right-of-first-refusal.2 The Rapoports responded on July 11, 1997, rejecting Mayes’s offer on the basis that it did not match the Key contract of June 13, 1997. Despite the rejection, the Rapoports continued to negotiate with Mayes for the next two months.3
On August 22, 1997, Mayes notified the Rapoports that Mayes was “prepared to go to settlement under the same terms and conditions as those offered to [Key].” The Rapoports replied that Mayes would have to obtain financing on her own for the full purchase price and would have to be prepared to close the sale of the property on August 25, 1997. Mayes did not agree, and on November 18, 1997, the Rapoports closed their sale of the property to Key.
B.
Shortly after Mayes sued in state court, the Rapoports removed the case to the District of Maryland on the basis of diversity.4 Thereafter, but before the Rapo-*460ports filed their answer to Mayes’s complaint, Mayes amended her complaint to add Key as a defendant — a significant addition because Key is a resident of Maryland and his addition as a defendant seemingly defeats diversity jurisdiction.
On June B, 1998, a few months after Mayes filed her amended complaint, the district court identified this issue and requested that the parties brief the question of continued federal jurisdiction.5 Following briefing, by its ruling of October 13, 1998, the district court retained jurisdiction over the case:
The court is aware that both plaintiff Mayes and defendant Key are residents of Maryland, seemingly eliminating jurisdiction in this court on the basis of diversity of citizenship. However, where the defendant demonstrates that the plaintiff “cannot establish a claim against the nondiverse defendant even after resolving all issues of fact and law in the plaintiffs favor,” the joinder of such a party is deemed fraudulent and does not defeat diversity for federal jurisdictional purposes.
As the analysis in this Memorandum indicates, Mayes has failed to establish a claim against Key. As a result, consideration of the merits of the Rapoports’ motion to dismiss is appropriate.
J.A. 199-200 (citations omitted). The district court went on to hold that Mayes could not state a claim against the Rapo-ports, and it dismissed the case against each of the defendants.
Mayes has appealed the dismissal of her case, and we possess jurisdiction pursuant to 28 U.S.C. § 1291.
II.
We review de novo questions of subject matter jurisdiction, including those relating to the propriety of removal and “fraudulent joinder.” See Mansfield, Coldwater & Lake Michigan Ry. Co. v. Swan, 111 U.S. 379, 384, 4 S.Ct. 510, 28 L.Ed. 462 (1884). Similarly, we review de novo a district court’s Rule 12(b)(6) dismissal. Estate Constr. Co. v. Miller & Smith Holding Co., Inc., 14 F.3d 213, 217 (4th Cir.1994). In reviewing a 12(b)(6) dismissal, we construe factual allegations in the nonmoving party’s favor, treating them as true, and “we will affirm a dismissal for failure to state a claim only if it appears that ‘the plaintiffs would not be entitled to relief under any facts which could be proved in support of their claim.’ ” Id. at 218 (quoting Schatz v. Rosenberg, 943 F.2d 485, 489 (4th Cir.1991)).
III.
A.
We begin, as we must in a diversity case, by examining the basis for jurisdiction. The district court relied upon decisions applying the doctrine of fraudulent joinder in order to retain jurisdiction in this diversity case. J.A. 200 (citing Marshall v. Manville Sales Corp., 6 F.3d 229, 232-33 (4th Cir.1993), and Kimmons v. IMC Fertilizer, Inc., 844 F.Supp. 738, 739 (M.D.Fla.1994)). However, there is authority that, after a case has been removed, federal courts may not rely upon that doctrine to justify jurisdiction. See Cobb v. Delta Exports, Inc., 186 F.3d 675, 677 (5th Cir.1999) (“The fraudulent joinder doctrine does not apply to joinders that occur after an action is removed.”) (emphasis in original); see also Gum v. General Electric Co., 5 F.Supp.2d 412, 415 n. 8 (S.D.W.Va.1998) (noting that fraudulent joinder doctrine applies to defendants *461named before removal); Harrell v. Pineland Plantation, Ltd., 914 F.Supp. 119, 120 n. 4 (D.S.C.1996) (same). We therefore find it necessary, as a preliminary matter, to determine whether the fraudulent joinder doctrine is applicable to post-removal joinder. To make this determination, we analyze the legal framework relating to joinder of parties in removed cases, beginning our review with pre-removal joinder and then analyzing the procedures applicable after the case has been properly removed.
1.
Before a case has been removed to federal court, there are several rules that govern the ability of defendants to consummate removal. For example, the “complete diversity” rule clarifies that the statute authorizing diversity jurisdiction over civil actions between a citizen of a state where the suit is brought and a citizen of another state permits jurisdiction only when no party shares common citizenship with any party on the other side. Strawbridge v. Curtiss, 7 U.S. (3 Cranch) 267, 2 L.Ed. 435 (1806). This “complete diversity” rule, when coupled with other rules,6 makes it difficult for a defendant to remove a case if a nondiverse7 defendant has been party to the suit prior to removal.
The “fraudulent joinder” doctrine8 permits removal when a non-diverse party is (or has been) a defendant in the case. See Poulos v. Naas Foods, Inc., 959 F.2d 69 (7th Cir.1992) (cited in Marshall, 6 F.3d at 233); Triggs v. John Crump Toyota, Inc., 154 F.3d 1284, 1287 (11th Cir.1998) (noting that “[fjraudulent joinder is a judicially created doctrine that provides an exception to the requirement of complete diversity”). Under this doctrine, a district court can assume jurisdiction over a case even if, inter alia9 there are nondiverse named defendants at the time the case is removed. Marshall, 6 F.3d at 232-33; Cobb, 186 F.3d at 677. This doctrine effectively permits a district court to disregard, for jurisdictional purposes, the citizenship of certain nondiverse defendants, assume jurisdiction over a case, dismiss the nondi-verse defendants, and thereby retain jurisdiction. Cobb, 186 F.3d at 677-78. Since the fraudulent joinder doctrine justifies a federal court’s initial assumption of diversity jurisdiction, it has no effect once the district court actually possesses jurisdiction-including after the case has been removed.
2.
When a plaintiff seeks to join a nondiverse defendant after the case has been removed, the district court’s analysis begins with 28 U.S.C. § 1447(e), which provides the district court with two options: “If after removal the plaintiff seeks to join additional defendants whose joinder *462would destroy subject matter jurisdiction, the court may deny joinder, or permit joinder and remand the action to the State court.” These are the only two options for a district court faced with a post-removal attempt to join a nondiverse defendant;10 the statute does not allow a district court to retain jurisdiction once it permits a nondiverse defendant to be joined in the case.11
Under Section 1447(e), the actual decision on whether or not to permit joinder of a defendant under these circumstances is committed to the sound discretion of the district court; thus, this decision is not controlled by a Rule 19 analysis. See supra note 11; 14C Charles A. Wright, Arthur R. Miller, & Edward H. Cooper, Federal Practice and Procedure § 3739, at 445 (3d ed. 1998) (“Section 1447(e) gives the court more flexibility than a strict Rule 19 analysis”). In exercising its discretion under Section 1447(e), the district court was entitled to consider all relevant factors, including: “the extent to which the purpose of the amendment is to defeat federal jurisdiction, whether the plaintiff has been dilatory in asking for amendment, whether the plaintiff will be significantly injured if amendment is not allowed, and any other factors bearing on the equities.” See *463Gum, 5 F.Supp.2d at 414 (quoting Coley v. Dragon Ltd., 138 F.R.D. 460, 465 (E.D.Va.1990) (citing Hensgens, 833 F.2d at 1182)). The district court, with input from the parties, should balance the equities in deciding whether the plaintiff should be permitted to join a nondiverse defendant. Id.
As noted, some courts have held the doctrine of fraudulent joinder inapplicable after the case has been removed. See supra at 460-61. Those courts reason that, since the federal court already possesses jurisdiction, it need not “ignore” the citizenship of “fraudulently joined” defendants in order to dismiss them from the case; rather, it can simply decline to permit joinder of the nondiverse defendant in the first place. See Cobb, 186 F.3d at 678.12 We agree that the doctrine does not directly apply after removal because the district court already possesses jurisdiction. However, if the defendants can carry the heavy burden of proving fraudulent joinder, that fact should be a factor — and perhaps the dispositive factor — that the court considers in deciding whether a plaintiff may join a nondiverse defendant. Thus, the fraudulent joinder doctrine can be yet another element of the district court’s “flexible, broad discretionary approach” to resolving a post removal question of whether a nondiverse defendant should be joined under Section 1447(e). Gum, 5 F.Supp.2d at 414.
We previously applied this approach in AIDS Counseling and Testing Centers v. Group W Television, Inc., 903 F.2d 1000 (4th Cir.1990). There, the plaintiffs sought leave to amend their complaint in order to join a nondiverse defendant after removal. Id. at 1003. The district court denied the motion, believing that the plaintiffs “sought to amend their complaint solely to defeat diversity and to deprive the court of jurisdiction.” Id. Based on the facts present there, we held that it was not an abuse of discretion to decline permission to amend to add the nondiverse defendant. Significantly, notwithstanding that the plaintiff in that case sought to join the defendant after removal, we relied, in part, upon the doctrine of fraudulent join-der in affirming the district court’s decision. Id. at 1003-04.
3.
We emphasize that the district court was correct to carefully scrutinize Mayes’s attempt to add a nondiverse defendant after removal. Especially where, as here, a plaintiff seeks to add a nondiverse defendant immediately after removal but before any additional discovery has taken place, district courts should be wary that the amendment sought is for the specific purpose of avoiding federal jurisdiction. See AIDS Counseling and Testing Centers, 903 F.2d at 1003 (noting fact that “plaintiffs had filed their motion to amend shortly after the case was removed to federal court and before they had undertaken any discovery” supported denial of motion to amend); Gum, 5 F.Supp.2d at 415 (same). Careful scrutiny of attempts at post-removal, non-diverse joinder protects the diverse defendant’s “interest in keeping the action in federal court.” See Coley, 138 F.R.D. at 465 (citations omitted).
There are, however, other interests at stake when such a joinder is sought, including the “danger of parallel lawsuits in federal and state court, which may spawn inconsistent results and inefficient use of judicial resources.” Id. In this instance, it *464is apparent that the district court determined fraudulent joinder to be dispositive of the joinder question, inasmuch as it articulated no other basis for its dismissal of Key. Thus, having concluded that the district court could properly consider whether Key had been fraudulently joined by Mayes, we must turn to the application of the fraudulent joinder doctrine to this case.
B.
The central inquiry for jurisdictional purposes is whether the Rapoports, the removing parties, established that Mayes had fraudulently joined Key as a defendant. We have previously made clear that:
In order to establish that a nondiverse defendant has been fraudulently joined, the removing party must establish either: [tjhat there is no possibility that the plaintiff would be able to establish a cause of action against the in-state defendant in state court; or [tjhat there has been outright fraud in the plaintiffs pleading of jurisdictional facts.
Marshall, 6 F.3d at 232 (quoting B., Inc. v. Miller Brewing Co., 663 F.2d 545, 549 (5th Cir.1981)) (alteration in original; quotation omitted). “The burden on the defendant claiming fraudulent joinder is heavy: the defendant must show that the plaintiff cannot establish a claim against the nondi-verse defendant even after resolving all issues of fact and law in the plaintiffs favor.” Marshall, 6 F.3d at 232-33 (citing Poulos, 959 F.2d at 73). We recently noted that “[tjhis standard is even more favorable to the plaintiff than the standard for ruling on a motion to dismiss under Fed.R.Civ.P. 12(b)(6).” Hartley v. CSX Transp., Inc., 187 F.3d 422, 424 (4th Cir.1999). Further, in determining “whether an attempted joinder is fraudulent, the court is not bound by the allegations of the pleadings, but may instead ‘consider the entire record, and determine the basis of joinder by any means available.’ ” AIDS Counseling and Testing Centers, 903 F.2d at 1004 (quoting Dodd v. Fawcett Publications, Inc., 329 F.2d 82, 85 (10th Cir.1964)).
Here, the Rapoports do not assert “outright fraud”; rather, they argue that Key’s joinder does not defeat diversity because there is no possibility that Mayes can state a claim against Key. The district court agreed, holding that Mayes had extinguished her right-of-first-refusal by submitting a non-matching offer when she first attempted to exercise that right. In the alternative, the district court held that even if the right-of-first-refusal had not been extinguished by the initial non-matching offer, it was extinguished when Mayes failed to exercise the right in a timely manner. Based on these alternative conclusions that the right-of-first-refusal had been extinguished, the court held that Mayes could not state a claim against either the Rapoports or Key. Consequently, the court concluded that Mayes fraudulently joined Key as a defendant, dismissed Key, and retained jurisdiction.
To the contrary, we hold that Mayes in fact stated a claim against Key that withstands challenge on the basis of fraudulent joinder. By her amended complaint, Mayes claimed (1) that the Rapo-ports entered into a binding contract with Key, thereby breaching Mayes’s right-of-first-refusal by making it “impossible” for Mayes to exercise that right; (2) that Mayes would have matched Key’s offer if she had been given the opportunity to do so; (3) that the Rapoports thus improperly sold the property to Key; and (4) that Mayes is entitled to specific performance, ie., to get the property itself from Key. Put simply, Mayes’s claim for specific performance against Key requires that she prove (1) that Mayes’s right-of-first-refusal was breached; and (2) that she is entitled to the remedy of specific performance against Key.
1.
To determine whether the district court properly concluded that Key was fraudu*465lently joined, we must first evaluate — resolving all issues of fact and law in Mayes’s favor — her claim that her right-of-first-refusal was breached. Under Mayes’s version of the facts,13 Mayes possessed a “right of first refusal to match an offer to purchase the property.” J.A. 6 (emphasis added). Mayes alleges, and the Rapoports acknowledge, that, prior to accepting Key’s offer, they did not present the offer to Mayes. Instead, the Rapoports signed and accepted Key’s offer, creating a contract with Key. They then presented the Key contract for Mayes to match. The Key contract was not expressly conditional; in other words, it did not condition sale of the property upon, or even mention, Mayes’s right-of-first-refusal. Since the plain language of the right-of-first-refusal provided that Mayes had the right to match an offer, Mayes can and does state a claim; ie., that the Rapoports violated the plain terms of the right-of-first-refusal when they presented Mayes with the signed, unconditional Key contract, instead of an offer for the sale of the property.
The Rapoports argue, however, that it is “irrelevant” whether they presented Mayes with a contract or an “offer” because Mayes failed to match Key’s terms once they were presented. We disagree. There are relevant substantive differences between an offer and a signed, unconditional contract, and presenting Mayes with a signed contract, rather than an offer, could have altered Mayes’s responses in the exercise of her right-of-first-refusal.
For example, by attempting to match the Key contract, Mayes could have been buying herself a lawsuit. Among other things, even if Mayes had sent a “matching” offer to the Rapoports, the Rapoports had accepted the matching offer, and the sale to Mayes had closed, Mayes still could have been forced to sue to obtain equitable title14 from Key. Further, when she received the Key contract, Mayes was forced to guess whether the Rapoports would risk breaching it in order to attempt to comply with her right-of-first-refusal. Thus, the Rapoports’ failure to send Mayes an offer not only violated the plain terms of the right-of-first-refusal, but also could have had practical consequences on the value of the property and how Mayes decided to “match.” Under these circumstances, the Rapoports cannot credibly argue that presenting Mayes with the Key contract— instead of an offer — was irrelevant or harmless.
The Rapoports also argue on appeal that the accepted practice in Maryland is for a seller to accept an offer, creating a contract, and thereafter present that signed contract to the holder of a right-of-first-refusal. In substance, the Rapoports’ argument is that, in Maryland, there is no difference between presenting an offer or a signed contract to an option holder (ie., the person holding a right-of-first-refusal). However, in determining whether Key was fraudulently joined, we must resolve all issues of law in favor of Mayes, Hartley, 187 F.3d at 424; Marshall, 6 F.3d at 232-33, and the law of Maryland distinguishes between an offer and a contract. See Lemlich v. Bd. of *466Trustees of Harford Community College, 282 Md. 495, 385 A.2d 1185, 1189 (1978) (“It is so basic a contract principle that minimal supporting authority is needed to authorize the statement that there must exist an offer by one party and an unconditional acceptance of that precise offer by the other, prior to withdrawal by the offer- or, before a binding [contract] is born.”).
Further, the single decision that the Ra-poports rely upon to support their assertion that there is no difference in Maryland between presentation, to an option holder, of a contract and an offer, Yorkridge Serv. Corp. v. Boring, 38 Md.App. 624, 382 A.2d 343 (1978), does not support them. In that case, the seller notified the option holder that it had received an offer and gave the option holder a chance to match that offer. Yorkridge was not a case in which the seller presented a signed contract to an option holder. Thus, under the liberal fraudulent joinder standard, we are confident that Mayes has stated a claim for relief as to the first part of her claim against Key: that the right-of-first-refusal held by Mayes under the lease was breached.
2.
If the Rapoports breached Mayes’s right-of-first-refusal, then there is a possibility that she can obtain specific performance against Key. “[W]here a contract for the sale of realty is fair, reasonable and certain, it is as much a matter of course for a court of equity to decree specific performance as it is for a court of law to award damages for its breach.” Manning v. Potomac Elec. Power Co., 230 Md. 415, 187 A.2d 468, 472-73 (1963). A Maryland state court could well order specific performance, depending on, inter alia, Key’s knowledge and involvement in the breach. Although there may be a substantial burden on Mayes to establish entitlement to specific performance, “there need be only a slight possibility of a right to relief’ to defeat a claim of fraudulent joinder. Hartley, 187 F.3d at 426. We believe, at the very least, there is, in the Hartley sense, a “glimmer of hope” for such relief in this case.15 Id.
IV.
This is a dispute involving Maryland contract and real property legal principles, and we express no opinion as to any party’s likelihood of success on the merits. However, the Rapoports have not carried the heavy burden required to establish that Mayes fraudulently joined Key as a defendant. We therefore vacate the district court’s dismissal under Rule 12(b)(6), reverse its decision on jurisdiction, and remand with instructions to further remand to the state court.
VACATED IN PART, REVERSED IN PART, AND REMANDED WITH INSTRUCTIONS