There’s a new twist in the ongoing battle over whether a virus is linked to chronic fatigue syndrome (CFS). After the journal held it for 2 months, a study supporting a link between a mouse retrovirus and CFS was published today in the Proceedings of the National Academy of Science (PNAS). Many are still doubtful of the link, but they’re impressed by the authors’ efforts to ensure accuracy.

 In the new study, conducted by scientists at the National Institutes of Health (NIH), the U.S. Food and Drug Administration (FDA), and Harvard University, researchers scanned for traces of a virus known as XMRV in samples taken from 37 CFS patients, collected by Harvard Medical School CFS specialist Anthony Komaroff in the mid-1990s. They found evidence for the virus in 32 (87%) of the patients, but in only three out of 44 healthy controls (6.8%). It remains to be seen whether the infection causes the disease or vice versa, says NIH virologist and co-author Harvey Alter—but he’s “confident” that the findings are correct.

 XMRV—less succinctly known as xenotropic murine leukemia virus-related virus—was first implicated for its potential involvement in prostate cancer, a link that’s still under intense debate. Then, in a Science paper published last year, a team led by retrovirologist Judy Mikovits of the Whittemore Peterson Institute for Neuro-Immune Disease (WPI) in Reno, Nevada, found evidence of infection in 67% of CFS patients, compared with just 3.4% of healthy controls. But since then, four other papers failed to find the link, or any evidence of XMRV infection in humans at all. The last of the four, by researchers at the U.S. Centers for Disease Control and Prevention (CDC), was also held for a while, at the researchers’ request, while they tried to figure out how government labs could come to such opposite conclusions. The CDC paper was eventually published on 1 July in Retrovirology.

 Skeptics were concerned that the XMRV Mikovits had found might be the result of contamination by mouse DNA in the lab. To address this, the new study’s first author—FDA virologist Shyh-Ching Lo—and his colleagues tested every positive sample for murine mitochondrial DNA. They found none.

 While the paper was on hold—also because of conflicts with other studies—the team ran additional checks that bolstered the data further, says Alter. “I felt we needed to do more to prove our case,” Alter says, in part because an additional, third reviewer, had looked at the paper at PNAS’s request. For instance, the researchers took fresh samples from eight of the patients and found that, 15 years on, they were still infected and that the virus had evolved, “just as we would expect from a retrovirus,” says Alter. The wait was “time well spent,” he adds.

 The data do seem solid, admits Steve Monroe, director of CDC’s Division of High-Consequence Pathogens and Pathology. “It’s simply a good paper,” adds Reinhard Kurth, the former director of the Robert Koch Institute in Germany, who helped test some of CDC’s samples and did not find the virus either. Alter—a widely respected virologist and winner of the Albert Lasker Award for Clinical Medical Research—”clearly knows what he is doing. They did everything correctly,” says Kurth, who nonetheless says he remains skeptical.

 So too does virologist Robin Weiss of Imperial College London (ICL), who says he’s seen too many instances of proposed new human retroviruses that fell apart on closer inspection, including one he reported in arthritis and lupus patients in 1999 that turned out to be an innocuous rabbit virus. (In a 40-page review that he co-authored in 2008, Weiss called such mishaps “human rumor viruses.”) “You can have a very good reputation and be very careful and still get it wrong,” Weiss says.

 Part of the problem, skeptics say, is that the researchers didn’t exactly replicate the Science paper. XMRV is a so-called xenotropic murine virus, which means it can no longer enter mouse cells but can infect cells of other species. (Murine means “from mice.”) The researchers in the PNAS paper say the viral sequences they find are more diverse than that and resemble more closely the so-called polytropic viruses, which is why they adopted the term MLV-related virus, for murine leukemia virus. “Let’s be clear: This is another virus. They did not confirm [Mikovits’s] results,” says retrovirologist Myra McClure of ICL, a co-author of one of the four negative studies.

 Still, “in the grand scheme of things,” the viral sequence found in the PNAS paper closely resembles those of XMRV, says Celia Witten, the director of FDA’s Office of Cellular, Tissue and Gene Therapies, who was not an author of the paper herself but spoke on Lo’s behalf. Witten adds that the data “support” the Science paper. Mikovits—who is “delighted” by the new paper—says the difference is not important. In as-yet-unpublished results, her group finds more genetic diversity in the virus as well, she says.

 Meanwhile, a working group coordinated by the National Heart, Lung, and Blood Institute (NHLBI) is coordinating an effort to answer the most baffling question: Why some labs find the virus in both patients and healthy people, and others find it in neither. Initially, some believed there might be geographical reasons, because the first three negative studies were all from Europe—but that theory seems unlikely after the CDC paper, whose patients were from Kansas and Georgia. Patient selection could play a role: Different studies have used different diagnostic and recruitment criteria. But even given this messiness, it’s hard to explain why four studies wouldn’t have included a single infected patient.

 The discordant results may also stem from subtle differences in handling the samples or performing the tests that would have led the four labs to miss the virus. But CDC’s Monroe says he’s confident that the lab can identify the virus. As part of the NHLBI program, researchers at FDA, CDC, WPI, and other labs have all blindly tested a panel of samples, some of them “spiked” with different amounts of the virus; all of them performed well. Further exchange of samples and reagents is now under way to understand where the differences came from. “They should be able to clear this up by Christmas,” says Kurth.

 Many of the main players in the controversy plan to attend a workshop organized by NIH on 7 and 8 September. Mikovits, who is on the scientific committee, says she has seen the abstracts of two presentations confirming her findings. “I think it will be fun,” she says.