WASHINGTON (AP) - It was a hot lead for detectives on a cold case. People suddenly were getting salmonella at a Minnesota restaurant more than 1,000 miles from the center of the nation's outbreak.

Not my tomatoes, protested the manager. He'd switched his supply to government-cleared fresh tomatoes and even canned ones. But a lot of his menu items had a raw jalapeno garnish sprinkled on top, and that turned out to be a critical clue in the two-month salmonella mystery.

On July 3, Minnesota e-mailed the feds. After tracing credit card receipts - to find what the restaurant's healthy customers didn't eat - there was good evidence that the jalapenos were sickening people. And, officials had a diagram tracing the pepper shipments all the way back to three farms in Mexico.

One of those farms shipped peppers through the same large warehouse in McAllen, Texas, where Food and Drug Administration inspectors weeks later would find a single contaminated Mexican-grown pepper being packed by a neighboring vendor.

How could Minnesota pinpoint hot peppers just days after discovering a cluster of sick residents, when federal investigators had spent weeks fruitlessly chasing tomatoes?

To be fair, "there was already some doubt about tomatoes causing this whole outbreak," cautioned Kirk Smith, foodborne disease chief at the Minnesota Department of Health.

And federal investigators say Minnesota's information came just as they were getting hints from two Texas restaurant clusters that jalapenos might play a role.

"Ours was the first that pointed specifically to jalapenos as an ingredient, not just the salsa," Smith said.

It's too soon to know if the Centers for Disease Control and Prevention improperly blamed tomatoes in early June, based on reports from the first people to fall ill in New Mexico and Texas.

"I don't think we can find fault yet," said University of Georgia food-safety expert Michael Doyle. "With tomatoes, if you looked at the initial case-control studies, they really came up high on the list."

The CDC didn't comment Wednesday.

At the FDA, food safety chief Dr. David Acheson told The Associated Press the system should be reviewed to see if it can be improved. "Did every part of this system work from one end to the other?" he asked. "I'm not saying it didn't, but I think one has to question that."

Regardless, the way Minnesota unraveled its own cases - speedily comparing the sick and the well and then racing to track food suppliers - offers lessons for a public health system grappling with how to handle increasingly complex outbreaks from tainted produce.

"We have got to put the appropriate perspective on this outbreak as to what went right and what went wrong so the kind of changes that are going to further foodborne disease (prevention) can be made," said Michael Osterholm, a University of Minnesota infectious disease specialist and frequent adviser to the government.

He fears the salmonella mystery may be the "swine flu of foodborne disease," and make federal health officials more reluctant to issue consumer warnings in future outbreaks unless they've found the smoking gun, an actual tainted food.

"That would be the worst legacy of this entire situation," Osterholm said.

Reports of the salmonella strain sickening hundreds elsewhere in the country began dribbling in to Minnesota's state health department on Monday June 23.

Minnesota's system is different from those of many states: Rather than county health departments initially checking outbreaks and reporting to headquarters, Smith's state office handles investigations from the beginning. By Thursday, with six cases reported, he had epidemiologists interviewing the sick: What did you eat in the few days before getting ill? Where?

By Sunday, two people had mentioned the same Twin Cities-area restaurant. Smith ordered that other patients be directly asked about that site. Monday morning, four more people fingered it - and by lunchtime, epidemiologist Erin Hedican was on the scene.

She quickly found seven more ill: employees who ate their own meals at the restaurant and started getting sick after the first customers had. Good to know: That meant the workers weren't the source.

With the manager, Hedican combed ingredients. Any new items added lately? New suppliers? She requested invoices from shipments just before June 14, the first known meal date of one of the sick, and started the hard push to get credit card receipts so she could learn what people who didn't fall ill had eaten.

By Tuesday morning, a garnish made of diced jalapenos and red peppers was topping a list of possible suspects.

"This is not like a sprig of parsley on the edge of your plate. This was sprinkled directly on almost every entree," Smith said.

Still, "a lot of people didn't notice the jalapenos," Smith said, while they were quick to mention tomatoes.

"Recall, that's what makes it tricky. That's why I wonder about all those initial cases" in other states, he added.

By Wednesday night, Smith's team had interviewed 13 sick people and 28 others who had eaten at the restaurant on the same days but stayed well. The sick were 46 times as likely to have eaten the garnish. The next morning, he alerted CDC and FDA.

Meanwhile, Ben Miller of the Minnesota Department of Agriculture, which regulates food suppliers, was pursuing those invoices. Miller knows traceback: He is credited with following contaminated lettuce blamed for a 2006 E. coli outbreak back to two suspect farms in California, before FDA singled out the culprit.

This time around, Miller knew his colleagues down the hall were suspicious of that garnish. He doubted a red pepper connection; they're used in far more restaurants than jalapenos.

The Twin Cities supplier that delivered to the restaurant led him to a larger distributor, also local. Miller whittled down shipment dates to between June 5 and 9. That distributor had bought from two sources: a shipper in California and another in McAllen, Texas, who in turn got the peppers from three farms in Mexico. Miller later ruled out one farm by further narrowing shipping dates; now he's waiting to hear from FDA if his Texas link panned out.

"A few phone calls and you can work it fairly quickly back to the grower," Miller said.

Federal officials had lots of questions for Minnesota as they matched that data with the clusters in Texas, the outbreak's center.

The Minnesota data "helped us begin to narrow this down," Acheson said, although he wouldn't call it the key cluster.

But Smith's team wasn't done: By July 8, it had a big enough group - 19 sick and 78 healthy customers - to do a statistical comparison of multiple ingredients. The sick were 100 times as likely to have eaten a jalapeno as the well.

The next day, July 9, the CDC issued its first consumer precaution, that people at high risk of salmonella should avoid fresh jalapenos.

Associated Press Writer Ricardo Alonso-Zaldivar contributed to this report.