When the biotech company Bayer CropScience AG requested federal permission in August to market a variety of gene-altered rice, it assured itself a small, unwanted place in history: the first to seek approval for a genetically engineered food that was already -- illegally -- on the market.

Now, as federal regulators consider that belated application, they are finding themselves under scrutiny, too -- from scientists and others who say the 20-year-old system of biotech crop oversight is failing.

The Bayer lapse is the latest in a string of problems, critics note, including taco shells and other foods contaminated in 2000 with unapproved StarLink corn, the accidental release in 2002 of crops engineered to make a pig diarrhea vaccine, and the growing prevalence of "superweeds" that have acquired biotech genes that make them impervious to weed killers.

Federal officials are still investigating how the experimental "LLRICE601" escaped from Bayer's test plots after the company dropped the project in 2001. When they announced 10 weeks ago that the unapproved variety had become widespread in the nation's long-grain-rice supply, countries around the world blocked imports from the United States, rice futures plummeted and hundreds of farmers sued Bayer.

Bayer's response -- a hasty application for government approval, expected to be granted within weeks -- has been greeted with concern by many agriculture experts who fear that the action, though likely to ease Bayer's legal woes, will make matters worse for farmers and the environment.

"Are we going to do this every time a new transgene that we didn't intend to get out gets out?" asked Norman Ellstrand, who directs the Biotechnology Impacts Center at the University of California at Riverside.

LL601 contains a bacterial gene that protects rice from Bayer's Liberty weed killer, allowing farmers to use the chemical without harming their crop. The prospect of widespread cultivation worries many experts, who say the key gene is sure to move via pollen into red rice, a weedy relative of white rice and the No. 1 plant pest for rice farmers in the South.

Thus endowed, red rice would become immune to the herbicide, increasing its economic havoc.

Experts point to other troubling elements of the Bayer petition.

Nearly 40 percent of its pages, for example, are blacked out as "CBI," or confidential business information, even though the approval process is by federal statute supposed to be public.

Also at issue is the regulatory shortcut that Bayer is using, which allows a company to skip many of the usual safety tests by claiming that the new variety is similar to ones already approved -- in this case, two approved varieties of biotech rice that Bayer never commercialized because farmers did not want it around their fields.

Bayer, with U.S. headquarters in Research Triangle Park, N.C., is adamant that LL601 poses no risk, and even critics generally agree that it is safe to eat. The bacterial gene that is in LL601 is also in several approved varieties of engineered corn, canola and cotton.

"We believe that our herbicide-tolerant rice would contribute significantly to rice productivity," said company spokesman Greg Coffey, adding that Bayer nevertheless has no immediate plans to market the product.

In a draft environmental assessment released with extraordinary rapidity last month, the Animal and Plant Health Inspection Service (APHIS), which handles biotech crop approvals for the Agriculture Department, announced a "preliminary decision" to approve -- or in agency parlance "deregulate" -- LL601.

Among those favoring approval is the USA Rice Federation, which represents many rice growers. The group has opposed introducing engineered rice to U.S. fields, but it is now more concerned about the European Union's ongoing refusal to buy American long-grain rice laced with LL601.

U.S. approval would not guarantee European acceptance. But it is "the best available response to a major commercial issue," the federation wrote to APHIS.

Many weed experts see the relative risks and benefits differently, however. They agree with APHIS and Bayer that cross-pollination between white rice and red rice is rare, probably occurring less than 1 percent of the time. But multiply that by millions and millions of rice plants, they say -- and then start using Liberty, which by killing conventional red rice will allow the resistant weed to dominate -- and within a few years, huge expanses of the South could be infested with Liberty-resistant red rice.

"Anyone who works with rice and red rice knows it," said Cynthia Sagers, a plant ecologist at the University of Arkansas. "It's going to happen."

The government's environmental assessment contends that farmers can fall back on other herbicides when that occurs, but opponents say that solution is shortsighted. They note that as gene-altered crops have become common -- some 70 varieties have been approved in the past 15 years, many of them engineered to be resistant to various weed killers -- it has become common to find weeds that are immune to two or even three weed killers.

"We have no ability to absolutely contain these things once they're grown outside," said Rene Van Acker, a weed ecologist at the University of Guelph in Canada.

Others are complaining that Bayer's application is effectively a secret document because of the material blacked out as confidential business information.

"It makes the public reliant on the interpretation of the data by Bayer, which is not a disinterested or unbiased party," wrote the Washington-based Center for Food Safety in comments to regulators.

Rebecca Bech, associate deputy administrator for biotechnology regulatory services at APHIS, defended the application, saying it is "fairly typical to have a lot" of redacted proprietary information in biotech crop applications.

But a review of the five most recently released applications submitted by companies, including ones for genetically engineered corn, grass, alfalfa and cotton, shows that four of those five had no such deletions. (The fifth notes that information has been deleted but does not say how many pages.)

Still others question the procedure Bayer is using to seek LL601 approval. Instead of going through a full deregulation process, it applied for an extension of approvals it won earlier for two other herbicide-resistant rice varieties developed nearly a decade ago.

That shortcut was created in 1997 to streamline approvals. But critics say the record of problems indicates a need for more careful oversight, not quicker approvals.

To allay concerns, Bayer has submitted with its application a "stewardship plan" -- voluntary farming practices, including extra dosings of Liberty, aimed at minimizing genetic crossovers to red rice.

Critics doubt that farmers will spend the extra time and money if they're not required to.

"Farmers are already under huge economic pressure," said Doug Gurian-Sherman, a senior scientist at the Union of Concerned Scientists. "It's just not going to happen."

But others, such as Johnny Saichuk, a rice specialist at the Louisiana State University AgCenter, support the approach.

"People are becoming better stewards," he said. "The sloppy managers who let it outcross will lose the technology. The good farmers will not have problems."

Even if LL601 is approved, Bayer's problems will not be over. It may be impossible to get every last seed of LL601 out of the U.S. long-grain-rice supply. And negotiations between American and European Union officials broke down last month over how much contaminating LL601 will be considered acceptable in exported rice.

The company also faces dozens of lawsuits, which may soon be combined into a large class action.

Reassuringly to Bayer, and infuriatingly to others, the trouble appears not to have weakened regulators' trust in the company.

Since learning of the contamination this summer, APHIS has received applications from Bayer to start field experiments on nine new kinds of gene-altered crops.

To date, eight of those have been given a green light.