The well-dressed woman in the waiting room was yellow, recalls John Hanson, MD -- a clear sign of jaundice. That was puzzling: One month earlier, in March 2005, Vivienne Wardley (not her real name), 51, had been in excellent shape except for a cold. A health-conscious woman, Wardley avoided prescription drugs and drank moderately. But tests showed that she needed an emergency liver transplant. And when Dr. Hanson, a gastroenterologist in Charlotte, North Carolina, examined her liver, he was startled. It was only a third of its normal size and showed massive tissue death.

What could have attacked Wardley's liver so quickly? The doctor remembered another patient his partner had treated in February. Usually healthy, Ramiro Pulquero, 26, walked into the Carolinas Medical Center in Charlotte jaundiced, feverish and vomiting blood. He died three days later. The autopsy showed that his liver, like Wardley's, had suffered tissue death.

Dr. Hanson soon learned that both patients had taken Ketek, a new antibiotic, for their minor respiratory infections. Strangely, they'd been on the pills for just five days before their livers failed. At that same hospital, another patient suffered liver damage after three days on Ketek; luckily, he recovered after stopping the drug.

Wanting to sound the alarm, Dr. Hanson and his colleagues published an online report in January 2006 in the Annals of Internal Medicine, a journal widely read by physicians. Dr. Hanson's phone started ringing the moment the story appeared. Most callers were doctors who'd seen similar cases. Another call was from the FDA: David Ross, MD, a medical reviewer who had tried, unsuccessfully, to stop Ketek's approval. "I thought there was a problem with this," he told Dr. Hanson. "I'm really glad you wrote this paper."

Dr. Ross knew something that Dr. Hanson didn't. "The company submitted fabricated data, the FDA knew it, and they approved the drug anyway," Dr. Ross says. Many of the patients in Ketek's clinical trial, in fact, didn't exist -- a point an FDA higher-up had ordered him not to tell the agency panel that was considering whether to approve the drug.

"I don't believe spending time on these issues in front of the advisory committee will be productive," the supervisor wrote in a January 2003 e-mail.

The "issues" were mind-boggling. The physician who enrolled the most patients in the study, an Alabama weight-loss doctor, allegedly forged scores of signatures, enrolling "volunteers" every few minutes.

By the time of the FDA review, she was under criminal investigation. (She's now in federal prison.) Another key researcher had been put on probation by the California medical board for gross negligence. He was arrested shortly after the study ended, when police, called to his home on a domestic violence complaint, found him with a bag of cocaine and waving a loaded gun at imaginary people. The study was so riddled with fraud and error that FDA reviewers decided it was useless.

Yet Dr. Ross says he was told to divulge nothing about those problems to the advisory board, which recommended that the drug be approved. Later, he says, he was pressured to soften his report about Ketek's liver toxicity to give higher-ups, as he was told, "wiggle room" to okay it.

Six million Americans have now used the drug, including hundreds of infants in a clinical trial designed to test Ketek's effectiveness against ear infections. "How does one justify balancing the risk of fatal liver failure against one day less of ear pain?" one FDA scientist, Rosemary Johann-Liang, MD, protested -- to no avail -- in a memo to her superiors. Most ear infections clear up in a few days on their own, she says.

The agency says the controversy is overblown. "There was enough good, solid scientific data to make that decision," says FDA spokeswoman Julie Zawisza, pointing to what appeared to be a history of safe use of Ketek in other countries. Ketek has now been linked to 18 deaths and at least 134 cases of liver damage, according to an independent analysis using FDA data. The real toll, some researchers say, may be far greater.

In December 2006, after Dr. Ross stood up at an FDA advisory board hearing and told an astonished audience about the saga of the drug, the panel urged the agency to strictly limit its use. Last February, one day before a Congressional committee began probing the FDA's handling of Ketek, agency officials abruptly rescinded its use for bronchitis and sinusitis and placed a strong "black box" warning on its label. By that point, Ketek's maker, Sanofi-Aventis, had already suspended its study on ear infections.

Last October the FDA sent a warning letter to Sanofi-Aventis, accusing the company of knowingly presenting compromised data to the agency, a charge the company denies. "We were not aware of the fraud," says spokeswoman Melissa Feltmann. "It was not until the FDA's criminal investigators uncovered it that we became aware of it."

The question remains, What did the FDA and the drugmaker know about the fake safety data, and when?

Congressmen John Dingell and Bart Stupak, both Michigan Democrats, are investigating that mystery right now in Congressional hearings.

Luckily, Vivienne Wardley recovered. But "unfortunately," Stupak says, "the truth comes too late for some victims."