The court’s opinion is here. Though this is decided on state constitutional grounds (which insulates it to a great extent from U.S. Supreme Court review), the court’s decision has the same structure of those other court decisions that have struck down similar laws in other jurisdictions. (1) These laws impose too great a burden on the right to vote, and indeed can constitute a poll tax to the extent that the state does not pay the expenses associated with obtaining the documentation necessary to get the i.d. and (2) the state has not shown a sufficient interest in the law as a means of fraud prevention, given the paucity of evidence that such laws would meaningfully prevent “impersonation” voter fraud.
Footnote 28 is especially interesting for those who would draw parallels to the state’s interest in limiting contributions to candidates in the names of preventing the appearance of corruption:

Appellants cite to First Amendment cases addressing campaign finance restrictions to support their contention that the perception of fraud or corruption should be entitled to greater weight. See, e.g., McConnell v. Federal Election Comm’n, 540 U.S. 93, 143 (2003) (“the prevention of corruption or its appearance constitutes a sufficiently important interest”). In the context of campaign finance reform, the appearance of corruption arises directly from the extensive financial contributions made to political candidates by those with a stake in legislative decisions. The statutes under review in campaign finance cases are all narrowly tailored to address and limit those contributions. Even though the United States Supreme Court has recognized the importance of combating the appearance of corruption, it has nonetheless invalidated many of these statutes precisely because they impose a severe and undue burden on fundamental rights under the First Amendment. See, e.g., Randall v. Sorrell, — U.S. —, 126 S. Ct. 2479, 2486 (2006) (holding that campaign finance restrictions are unconstitutional because “they impose burdens upon First Amendment interests that . . . are disproportionately severe”). Unlike the campaign finance laws, the Photo-ID Requirement does not address any perception of voter fraud with precision, nor is it necessary to solve the existing voter fraud problems. On these facts, perceptions alone are insufficient to justify substantial burdens on fundamental rights.

One state court justice dissented, and pointed to this evidence of impersonation voter fraud in the record:

Although I would not reach the merits of the claim against the permanent provisions of the MVPA due to lack of ripeness, I cannot leave unchallenged the majority’s incomplete recitation of the facts pertaining to the existence of voter fraud and the need for a photo ID system to combat that fraud. According to the majority, there has been no fraud in the polling places; thus no need to prevent it. But the evidence, in part, is this: In an investigative report issued after the 2000 presidential election by outgoing Secretary of State Rebecca McDowell Cook, and introduced in evidence in this case, “135 people who were not registered to vote were permitted to vote at a polling place without a court order and without apparent authorization from [an election] Board Official.” A subsequent report from then Secretary of State Matt Blunt noted, as even the plaintiffs have acknowledged here, that 79 voters registered from vacant lots, 45 people voted twice, and 14 votes were cast by the “dead.” Further, as part of a federal investigation in the aftermath of the 2000 election, the United States Department of Justice found a stunningly large number of duplicate and ineligible voter registrations throughout the state. According to that report,
[A] comparison of State voter registration data posted on the website of the Missouri Secretary of State with data from the United States Census Bureau indicates that at least 34 (nearly one-third) of the election jurisdictions in Missouri had more registered voters in November 2004 than there were persons of voting age in those jurisdictions under July 2003 Census estimates (released September 2004), and that 29 election jurisdictions in the State had more registered voters in November 2004 than there were persons of voting age in those jurisdictions under July 2004 Census estimates (released August 2005). Indeed, the State’s data indicates that the local election jurisdiction with the highest ratio, Reynolds County, had 153% of its 2003 Census voting age population, and 151% of its 2004 Census voting age population, registered to vote in the November 2004 federal election. This State’s data further indicates that, statewide, Missouri had voter registration totals in November 2004 amounting to 98 percent of the state’s voting age population according to July 2003 Census estimates and 96 percent of the state’s voting age population according to July 2004 Census estimates.
Although the majority agrees that there is some evidence of voter fraud at the voter registration stage, they discount that evidence as if it had no connection with fraud at the polling place. But why else does voter registration fraud occur if not to vote persons fraudulently registered? And if, as in the DOJ report, there are more voters registered to vote than persons eligible to vote, the requirement to present a photo ID will at least eliminate those who attempt to vote in the place of others and those who attempt to vote more than once. It must be said, too, that even if there were no substantial evidence of existing voter impersonation fraud, legislatures are permitted to respond to the potential for such fraud, and they may do so “with foresight” rather than “reactively.” Munro v. Socialist Workers Party, 479 U.S. 189 (1986). In any event, as the Carter-Baker Commission recently concluded, “there is no doubt that [in-person voter fraud] occurs” and that such fraud “could offset the outcome of close elections””

Why else would registration fraud occur? Because there is sometimes a bounty paid for each registration turned into election officials. The more important question is whether these number, assuming they are accurate (they may not be) would justify an onerous voter identification law. Spencer Overton’s recent book and forthcoming article explore this question very nicely.
More commentary from Bob Bauer.