For beef lovers, 2007 will go down as another year of eating dangerously.

Since the spring, meat suppliers have recalled more than 30 million pounds of ground beef contaminated with the potentially lethal bacteria E. coli O157:H7, including the 21.7 million pounds recalled by New Jersey-based Topps Meat in September.

After three relatively quiet years, the 20 recalls this year have raised new doubts about whether the beef industry's attempts to keep the pathogen out of ground beef, and the government's oversight of those efforts, are working.

Agriculture Department officials, who oversee the safety of pork, beef and poultry, say they did not recognize that anything was seriously amiss with the beef supply until the Topps recall hit.

Microbiologists say the prevalence of E. coli O157:H7 in the environment is highly variable, and no one can say with certainty what caused the spike in outbreaks. In several instances this year, however, USDA officials missed red flags and were slow to correct longstanding deficiencies in the way they monitor beef processors' efforts to contain the pathogen.

USDA officials did not learn that Topps had begun testing its ground beef less frequently until the recall. Recurring sanitation problems at a United Food Group plant in Vernon, Calif., that later recalled 75,000 pounds of ground beef did not trigger further enforcement actions because the agency had not told inspectors what to do about repeat violations. The recall was eventually expanded to 5.7 million pounds. Critics said the agency missed an opportunity to strengthen its early-warning system by not keeping track of every instance when a plant found the dangerous strain of E. coli in raw ground beef.

The department has postponed plans to target inspections at plants that had a record of problems because officials do not know which plants pose the greatest risks.

Similar lapses have surfaced during the seven years since meat processors were required to come up with scientifically based plans to contain and control pathogens. In 2002, USDA officials did not know that the E. coli strain had been detected in ground beef at ConAgra's Greeley, Colo., plant 63 times in the weeks leading up to a massive recall. The agency had been testing for the bacteria in raw ground beef since 1994, but skipped ConAgra's plants under a policy that exempted the largest processors. USDA now tests ground beef at every plant at least once month, while self-testing at plants remains voluntary.

E. coli O157:H7 is a variant of the bacteria normally found in animal and human intestines, and it spreads easily among cattle. Surveys of feedlots have shown that in the summer 63 to 100 percent of cattle could be infected, according to the Centers for Disease Control and Prevention. The bacteria is shed in feces and can contaminate meat during the slaughtering process.

For years, USDA testing showed the bacteria's prevalence in raw ground beef was increasing. Then it dropped by more than 80 percent between 2000 and 2005. The number of people who got sick also began to fall. In 2006, the CDC reported a 29 percent decrease in illnesses caused by E. coli O157:H7, compared with a baseline established from 1996 to 1998.

The beef industry and federal regulators thought such industry interventions as steam vacuums, pasteurization and acid washes were keeping the bacteria off meat. They were reluctant to question their presumed success, even as cases of human illness began rising over the past two years. When asked by lawmakers about the uptick in illness at a hearing in April, USDA officials said contaminated produce was the likely culprit. The day after the hearing, the USDA announced two beef recalls that were prompted by people becoming sick.

By the time Americans fired up their grills for the Fourth of July, cases of human illness had prompted five beef recalls. "We began to believe we may have a problem, but we weren't sure," Richard Raymond, undersecretary of agriculture for food safety, said in an interview last week. At that point, however, the agency "wasn't ready to make sweeping changes."

Then the Topps recall occurred. Raymond called it a "wake-up call."

The recalls have Carol Tucker Foreman, a former assistant secretary of agriculture who is now with the Consumer Federation of America, questioning whether regulators and the industry ever had a handle on O157:H7.

"I had assumed the steps the companies are taking were effective," she said. "Now I don't know if the falloff during the past several years was the result of the steps the industry took or whether we had a period of time where there wasn't much E. coli."

Stan Painter, a USDA inspector and representative of the inspectors' union, said not much has changed since the ConAgra recall.

"We're relying totally on the plant. We're doing very little testing ourselves," Painter said. "We're saying, 'You tell us you have a problem. And if we don't hear from you, we assume you don't have a problem.' "

On Monday, in a report requested by Rep. Rosa DeLauro (D-Conn.), the USDA inspector general said the agency lacked the data, management controls and technology to identify the plants at greatest risk for contamination. The USDA has not assessed the food safety plans at all processing plants. At 15 facilities, the inspector general found no record that inspectors have been reviewing plant test results at least once a week. In the case of United Food Group, the inspector general said the plant had been cited several times for sanitation problems, but inspectors did not take further action because they had no guidance on how to treat repeat violations. The USDA was supposed to issue such guidance after the ConAgra recall.

Until September, the agency also chose to disregard samples of raw ground beef found to have O157:H7 if the processor agreed to cook it. The reasoning was that because the tainted meat was not being sold raw, it did not pose a public health risk.

But ignoring those samples had some unintended consequences, said Felicia Nestor, a senior policy analyst with Food and Water Watch. The findings were far less likely to trigger a review of the plant's pathogen control practices and could not be used by USDA officials to identify trends.

"I believe we're seeing the results of that policy now," said Nestor, who called the USDA's methods "voodoo science."

Randall Huffman, vice president for scientific affairs of the beef industry's American Meat Institute Foundation, defended the USDA's sampling methods as accurate.

"The finished product is a reflection of the finished food safety system. Random testing is . . . the best measure of how well food safety works. The arguments that the data was skewed are absolutely false," he said.

USDA scientists, however, were persuaded that their data could use improvement. This fall, the agency said it would begin testing samples even if the meat had been diverted to cooking. The results would be compiled in a separate database.

In November, the agency required all plants to verify that their safety plans were working to contain O157:H7. Next month, it will begin testing imported trim -- the meat left after quality cuts are removed. Its a significant development because processors are increasingly buying trim from suppliers overseas. Canadian trim turned out to be the source of contamination at Topps. For the first time, it will also look at corporate practices to see whether there is a pattern of violations at multiple plants, FSIS spokeswoman Amanda Eamich said. The inspector general is also reviewing the FSIS E. coli testing programs.

Raymond said he welcomed the scrutiny. "Any time you have somebody from outside come in and take a look, it's always helpful," he said. "I didn't come here to supervise recalls. I came to prevent recalls."