A class action in Alabama cost Dexter Kamilewicz $91.33 in attorney fees to recover $2.19 on the merits. When he learned of this news, he and the other plaintiffs (his wife and Martha Preston) sued the class action attorneys (as well as certain defendants in the Alabama action) in the District Court for the Northern District of Illinois. There, to add to their chagrin, they ran up against another obstacle — the Rooker-Feldman doctrine. This is their appeal from an order dismissing their federal case for lack of subject matter jurisdiction.
Kamilewicz and the other plaintiffs have had mortgages serviced by BancBoston Mortgage Corporation, one of the defendants. They were among an estimated 715,-000 members of a class in a nationwide class action filed in the circuit court for Mobile County, Alabama. The suit—Hoffman, et al. v. BancBoston Mortgage Corp.—challenged the manner in which BancBoston calculated the amount of surplus each member of the class was required to maintain in their escrow accounts. Deposits to the escrow accounts were paid as part of the class members’ monthly mortgage payments. There was no question regarding the ultimate ownership of the surplus; it belonged to the mortgagor and was to be returned when the' mortgage debt was satisfied. The issue in the lawsuit was the propriety of BancBo-ston’s holding the surplus until the time it would be returned to the mortgagor.
In October 1993, the Alabama court granted partial summary judgment in favor of the plaintiff class. The court found that BancBo-ston’s practice was, in fact, inconsistent with the terms of the mortgages. Then in October 1995, counsel for the class prepared a notice of a proposed settlement of the suit. The notice stated that the settlement was “fair, reasonable, adequate, and in the best interests of the class” and that the attorney fees sought were “reasonable” and would “not exceed one-third of the economic benefit” to the class. Bank of Boston — one of the defendants — objected to the notice because it failed to advise the class that there were “substantial adverse effects” to the proposed settlement.
Those perceived adverse effects were that if BancBoston were ordered to refund the escrow surplus and if the attorneys for the class were to seek attorney fees out of the refund, some class members would suffer an out-of-pocket loss as a result of the lawsuit. It should be noted here that the Bank of Boston and BancBoston had offered, prior to the grant of partial summary judgment in 1993, to settle the litigation. As part of that proposed settlement, the bank and BancBo-ston would have had to pay $500,000 in attorney fees out of the banks’ own funds, and the entire amount of the escrow refund would have gone to the class members.
Nevertheless, the Alabama court approved the proposed notice and held a fairness hearing on January 10 and 11, 1994. On January 24, 1994, the court approved the settlement under which the class members received onetime interest payments ranging from $0.00 to $8.76. The court also found the attorney fees reasonable. The fee award was a percentage of the escrow accounts, or, the complaint in the present case asserts, in excess of $14 million. The fact may be that the award was closer to $8,500,000, but because in either event we are talking about signifi*509cant amounts of money, the actual amount is not material and, of course, the allegations of the complaint are accepted as true at this point in the litigation. Under the settlement, BancBoston deducted attorney fees from the class members’ escrow accounts. The deduction was recorded on the 1994 annual tax and interest statements as a “miscellaneous disbursement.” In most cases, the “miscellaneous disbursement” was more—far more— than the interest refund. It was this statement that, understandably, caught Mr. Kami-lewicz’s eye.
Unhappy with the peculiar result of the class action suit in Alabama, Mr. and Mrs. Kamilewicz and Ms. Preston filed the present federal class action against the bank and the mortgage company, and the plaintiffs’ attorneys in the Alabama action. The complaint contained allegations of violations of the Racketeer Influenced and Corrupt Organizations Act, 18 U.S.C. § 1962, and of the Civil Rights Act, 42 U.S.C. § 1983, as well as claims of common law fraud, negligent misrepresentation, attorney malpractice, breach of fiduciary duty, and conversion. In turn, the defendants in the federal case filed a motion with the Alabama court seeking an order directed to the Kamilewicz plaintiffs to show cause why they were not bound by the January 1994 order of that court. A hearing was scheduled in Alabama, to which the Kamilewicz plaintiffs sent an attorney for the purpose of stating that they would not participate in the hearing because they contested whether Alabama had personal jurisdiction over them. On January 30, 1996, the Alabama court reaffirmed the order of settlement.
Meanwhile, back in Chicago, on December 15, 1995, Judge Paul Plunkett of the Northern District of Illinois, 1995 WL 758422 had entered his order dismissing the federal case for lack of subject matter jurisdiction. He found that even though the federal case was “dressed up” as a claim for RICO damages or for attorney malpractice, it was, nevertheless, a collateral attack on a state court judgment which would require that he consider issues “inextricably intertwined” with the state court case. This he was prohibited to do, he concluded, under the doctrine known as Rooker-Feldmcm.
We review de novo his decision on a motion to dismiss the complaint, taking all well-pleaded allegations as true, and we draw all reasonable inferences in the plaintiffs’ favor. Deveraux v. City of Chicago, 14 F.3d 328 (7th Cir.1994); Rueth v. United States Environmental Protection Agency, 13 F.3d 227 (7th Cir.1993). Having performed those tasks, we find ourselves in agreement with the district court.
We first note that the exact ground on which the plaintiffs and, for that matter, the Amici, who are attorneys general from several states, stand in this appeal is a little like shifting sand. The plaintiffs argue that the judgment of the Alabama court is null and void because the court did not have personal jurisdiction over them and that the special protections required for class actions—notice, adequate representation, etc.—were not complied with. However, they also say that they are not seeking to overturn the Alabama judgment. Then, they state that where the district court went wrong was to use a state court judgment which was “null and void” as the basis for the Rooker-Feld-man bar. In addition, plaintiffs veer into a claim that the January 30, 1996, Alabama court order cannot be a basis for the Rooker-Feldman bar because there was no personal jurisdiction over them, and furthermore that fraud claims are somehow outside the Rook-er-Feldman bar. The Amici say that they do not want to establish an exception to the general rule that absent class members may be bound by a nationwide class action settlement, but that on the unusual—and egregious—facts of this ease, the Kamilewicz action should be allowed to proceed.
The Rooker-Feldman doctrine grows out of two Supreme Court cases: Rooker v. Fidelity Trust Co., 263 U.S. 413, 44 S.Ct. 149, 68 L.Ed. 362 (1923), and District of Columbia Court of Appeals v. Feldman, 460 U.S. 462, 103 S.Ct. 1303, 75 L.Ed.2d 206 (1983). At its core, the doctrine is a recognition of the principle that the inferior federal courts generally do not have the power to exercise appellate review over state court decisions. In Rooker, the Supreme Court *510stated that because the jurisdiction of the federal district courts was strictly original, those courts cannot reverse or modify a state court judgment — even if that judgment is wrong. Only the Supreme Court has that power. Feldman involved proceedings denying him a waiver of a bar admission rule which required that applicants have graduated from an approved law school. The first issue was whether the action constituted a judicial proceeding. The Court determined that it was in fact such a proceeding and that a district court “has no authority to review final judgment of a state court in judicial proceedings.” At 482, 103 S.Ct. at 1315.
It becomes necessary to our analysis, then, to determine whether what the federal court was asked to do in this ease was, essentially, review a decision of a state court. Our cases offer means by which that determination is made. We ask whether the federal plaintiff seeks to set aside a state court judgment or whether he is, in fact, presenting an independent claim. GASH Associates v. Village of Rosemont, Illinois, 995 F.2d 726 (7th Cir.1993). Put another way, if the injury which the federal plaintiff alleges resulted from the state court judgment itself, then Rooker-Feldman controls, and the lower federal courts lack jurisdiction over the claim. It does not matter that the state court judgment might be erroneous or even unconstitutional. Nor does it matter that the time for appeal to the United States Supreme Court may have passed. Garry v. Geils, 82 F.3d 1362 (7th Cir.1996).
In addition, the state and federal cases also do not need to be directly on point. The Court found a lack of jurisdiction in Feldman itself when the federal constitutional claims were “inextricably intertwined,” at 486, 103 S.Ct. at 1316-17, with the state court judgment. As we have explained, lower federal courts might be engaging in impermissible appellate review even when asked to entertain a claim that was not argued in the state court but was “inextricably intertwined” with the state court judgment. Ritter v. Ross, 992 F.2d 750 (7th Cir.1993), cert. denied, 510 U.S. 1046, 114 S.Ct. 694, 126 L.Ed.2d 661 (1994).
Our cases have attempted to make the analysis of this issue easier by offering a sometimes helpful tip. That tip is that often, if a federal plaintiff was the plaintiff in the state court action, the doctrine he must contend with is res judicata, not Rooker-Feldman. On the other hand, if the federal plaintiff was the defendant in state court, Rooker-Feldman is the applicable doctrine. Homola v. McNamara, 59 F.3d 647 (7th Cir.1995); Nesses v. Shepard, 68 F.3d 1003 (7th Cir.1995). The tip, however, is merely that, and allows exceptions. As we pointed out in Garry, both the Rooker and Feldman plaintiffs were also plaintiffs in the state judicial proceedings. Although it may be that the federal plaintiffs position as either a plaintiff or a defendant in .the state court proceeding will “usually coalesce with the souree-of-the-injury standard,” as we said in Garry, it may not always do so, and the important issue remains the source of the injury: the issue is whether the federal plaintiff is injured by the state court judgment or by a prior injury at the hands of the defendant.
How does all of this theory apply to the case before us? As a preliminary matter, we note that the Alabama court specifically found the settlement to be fair and approved the fees. In his order of January 24, 1994, the state judge found that the notice complied with all requirements of due process, that the consent decree was fair and reasonable, and that the attorney fees were reasonable.
Plaintiffs seem to be urging us to reach the conclusion that we can overturn the decision of the Alabama court insofar as it involved a determination that procedural due process was complied with. As support for this proposition, they rely on decisions which hold that a party can attack a default judgment against him if it is entered by a court without personal jurisdiction over him. We see significant differences between default judgments and the judgment under attack here. We reject the plaintiffs’ far-reaching proposed exception to the Rooker-Feldman doctrine.
Another argument which seems slightly off-center is that the January 30,1996, order, *511finding the allegations of fraud “baseless,” is not subject to a Rooker-Feldman bar because it was entered in the absence of personal jurisdiction over the plaintiffs and it was entered after the federal lawsuit was filed. This argument is tied to the claim that the plaintiffs were precluded from litigating their fraud claim in the state court, and thus were required to—or had a right to—litigate it in federal court.
The January 30 order itself is not material to the issues before us. It is not necessary to rely on it to determine that the federal lawsuit was properly dismissed. The fact is that the proper court for an assertion of fraud in the procurement of a judgment is the one which rendered the judgment. As we stated in Homola, at 650:
Relief in civil cases differs from that in criminal cases in two principal respects. First, the civil time limits are shorter (in federal litigation, for example, a claim of fraud on the court, Homola's theme, must be made within one year, Fed.R.Civ.P. 60(b)(3)). Second, the losing party cannot move to a new forum but must present his argument to the court that rendered the judgment. See Harris Trust & Savings Bank v. Ellis, 810 F.2d 700 (7th Cir.1987). Homola has tried to shift not only courts but also jurisdictions, asking a federal judge to award relief for wrongs committed in state court. Instead of saying that there was no constitutional flaw in the state proceedings, the district court should have refused to entertain the claims, sending Homola back where he belongs—the original courts.
Alabama has a procedure by which a litigant can assert an independent action for fraud upon the court within three years of the entry of the fraudulently induced judgment. Alabama Rule of Civil Procedure 60(b). In addition, the Alabama court retained continuing jurisdiction over the class action. It is not for the federal court to decide these issues.
We are dismayed by the plaintiffs’ reliance on Derrickson v. City of Danville, 845 F.2d 715 (7th Cir.1988), and In re VMS Securities Litigation, 156 F.R.D. 635 (N.D.Ill.1994), for any proposition regarding the Rooker-Feld-man doctrine or the fraud allegations. In Derrickson, a, federal class action settlement was approved. Then the state’s attorney for Vermillion County, Illinois, opened a criminal investigation. The question before us was whether claim preclusion barred the criminal proceedings in state court, not, obviously, a Rooker-Feldman bar. Plaintiffs, in a footnote, acknowledge a problem using this case as authority for their argument as “it involved a state action following federal approval of a consent decree_” We wholeheartedly agree. Similarly in VMS, the prior judgment was an approval of a settlement in a federal class action. The issues were whether certain subsequent state court actions could proceed. Use of either of these cases for the proposition that in the present suit, the Rooker-Feldman doctrine does not present a bar is disingenuous, at best.
The Kamilewiczes were class member/plaintiffs in the Alabama suit. The part of the judgment they are unhappy with is the approval of the settlement as to the fees to be paid to their attorneys—fees which were assessed against them. The district court concluded that the Kamilewicz plaintiffs were in the position of defendants in this aspect of the state court proceedings. That conclusion has some support in logic and bolsters a finding of a Rooker-Feldman bar. More important, however, is the fact that the plaintiffs’ injuries are a result of the state court judgment. Their claim in federal court is a multi-pronged attack on the approval of the settlement regarding the attorney fees issue. Regardless of which of the specific federal claims the district court were to consider, it would run directly into the state court finding, entered after a two-day fairness hearing—that the fees were reasonable. The federal claims are “inextricably intertwined” with the state court judgment, whether that judgment is right or wrong.
We won’t deny that the Alabama judgment seems questionable on the surface. How can it be right that a plaintiff should recover less than $10 and have to pay nearly $100 in fees? But how can we—or the district court—know that we would have ruled another way? To determine that, we would have to go through the steps already taken in the state court— *512precisely what lower federal courts are not allowed to do under Rooker-Feldman. And perhaps if we undertook that analysis we would find that some less tangible economic benefit to the class justified the settlement.
That the facts seem at least superficially egregious brings us to the arguments of Ami-ci that on facts such as these an exception of some undefined sort should be carved out, and that if the plaintiff class was actually in the position of defendants on the attorney fees issue, then even under Phillips Petroleum Co. v. Shutts, 472 U.S. 797, 105 S.Ct. 2965, 86 L.Ed.2d 628 (1985), perhaps personal jurisdiction was lacking in the Alabama court. We decline to take the step urged by the Amici.
In Shutts, the Court loosened somewhat the requirements for assertion of personal jurisdiction over members of a plaintiff class and determined that due process would be satisfied if there was notice, an opportunity to appear in person or by counsel, an opportunity to opt out of the class, and adequate representation. The rationale was, in part, that absent plaintiffs ordinarily do not face the same exposure as defendants. Amici argue that if on the attorney fees issue the Kamilewicz plaintiffs were actually in the position of defendants, then due process requires that issues of personal jurisdiction be assessed under the traditional minimum contacts analysis of International Shoe Co. v. Washington, 326 U.S. 310, 66 S.Ct. 154, 90 L.Ed. 95 (1945).
Even were we to accept the argument, it does not change the fact that lower federal courts cannot review decisions of the state court. Here, the state court approved the settlement, including the fees. The Supreme Court of Alabama or the United States Supreme Court could reverse the decision were either so inclined. The federal district court, on the other hand, cannot review the Alabama decision. We also note the obvious— that the posture of the Alabama case is not unique. In class actions, often the fees are a percentage of the award granted to the plaintiff class. The argument of Amici, if accepted, could have ramifications far beyond this case.
Because we agree that Rooker-Feldman bars this action, the judgment of the district court is Affirmed.