A chemist at the Food And Drug Administration (FDA) said testing in recent weeks had revealed some soft drinks contaminated with benzene at levels above the legal limit for water set by the US and Europe.
Benzene is listed as a poisonous chemical shown to increase the risk of leukaemia and other cancers.
The FDA was originally alerted in 1990 to the problem of benzene in soft drinks triggered by the preservative sodium benzoate. It never made the findings public, but came to an arrangement with the US soft drinks association that the industry would "get the word out".
But in recent months, internal documents and private tests have begun to surface, supported by claims from a former chemist for Cadbury Schweppes, who is now keen to blow the whistle on the health risk involved. He and a US lawyer commissioned new tests that have now prompted the FDA to re-open the case.
These independent tests, performed by a laboratory in New York, found benzene levels in a couple of soft drinks two-and-a-half-times and five times above the World Health Organisation limit for drinking water (10 parts per billion).
The FDA now confirms it has found a similar problem in its own follow-up testing. "There were a few isolated products that have elevated levels. We certainly want to make sure there is some reformulation," said an FDA chemist.
The problem is caused by two common ingredients - sodium benzoate and ascorbic acid (vitamin C) - which can react together to cause benzene formation. It is considered completely separate from other outbreaks of benzene contamination due to faulty packaging in the 1990s.
The two ingredients are still used together in a wide range of soft drinks across the world.
The FDA was first alerted to the problem in December 1990 by Cadbury Schweppes and Australian drinks group Koala Springs, according to an internal FDA memo.
This prompted FDA testing that led the US Department of Health and Human Services to report, again in an internal memo: "Benzene formation occurs at part per billion (ppb) levels in some food formulations containing sodium benzoate and ascorbic acid [vitamin C]."
These findings were discussed in a meeting between the FDA and National Soft Drinks Association (NSDA), representing the likes of Coca-Cola, PepsiCo and Cadbury Schweppes. The internal minutes from this meeting report that the companies, through the NSDA, "expressed concern about the presence of benzene traces in their products and the potential for adverse publicity associated with this problem".
But Greg Diachenko, an FDA chemist who helped to co-ordinate the FDA testing for benzene in 1990/91 and who took part in negotiations with the industry, said: "Soft drinks manufacturers told us [at the time] that they would get the word out and they were reformulating."
Legal action was not taken, he said, because the industry was already reformulating. This, he added, meant that any risk to consumers would be very short-term because risk analysis in the US works on the basis of lifetime exposure to a substance.
The FDA did more tests in the US in 1993 and found no problem. But, added Diachenko: "It is probable and likely that there were some people who did not get the message or that it was lost in the course of time."
Any soft drink companies founded since 1993, for example, could be outside of the loop and not know about the potential 'fixes'.
Plus, the FDA, as a US authority, only checked drinks available in the US. Diachenko said he was not sure how many other regulatory authorities had been told about the potential for benzene formation from sodium benzoate and ascorbic acid.
An article flagging up the problem was published in the public journal, the Journal of Agricultural and Food Chemistry, in 1993.
According to Diachenko at the FDA, if authorities outside the US had not seen the information they might not know, and that could mean more soft drinks containing benzene in their areas.
More than 1,500 soft drink products containing sodium benzoate and ascorbic acid or citric acid have been launched across Europe, Latin America and North America since January 2002.
Glen Lawrence, another chemist who conducted benzene testing for the FDA back in 1990-1991 and a co-author of the 1993 journal article, has also confirmed to BeverageDaily.com that sodium benzoate and ascorbic acid do react to form benzene in soft drinks.
His study showed that ascorbic acid initially reacted with metals, such as iron or copper, found in the water to create 'free radical' particles known as hydroxyl radicals.
Sodium benzoate, meanwhile, breaks down into benzoic acid when placed in acidic conditions, such as in a soft drink.
The hydroxyl radical attacks the benzoic acid, removing the carbon dioxide from it and leaving benzene in its wake. Lawrence's study said this reaction could take place "under conditions prevalent in many foods and beverages".
Lawrence said: "There is no good reason to add ascorbic acid (vitamin C) to soft drinks, and those that may have ascorbic acid naturally in them (juices) should not use sodium benzoate as a preservative. So it is really very easy to avoid the problem."
Sodium benzoate, also known as E211, is used as a preservative by a range of food and drink producers. Its main advantage is its effectiveness at killing off bacteria under the acidic conditions of most beverages.
Ascorbic acid, or vitamin C, is found naturally in fruit and vegetables but is also added as an antioxidant in food and drink production to help prevent spoilage and extend shelf-life.
The FDA says it has almost completed the testing. It has not yet made any public announcement about the problem, but has told BeverageDaily.com that it will soon address what action should now be taken.