Some brands of sodas were removed from the shelves over a concern that Sodium Benzoate may be converted to Benzene due to chemical reactions between other ingredients found in the sodas. As with most health risks, children are most vulnurable to the cancer risk associated with Benzene exposure. Pregnant women, who shouldn't be drinking soda in the first place, also put the developing fetus at a higher risk.

The concern is that the preservative Sodium Benzoate mixes with other common ingredients, and over time, changes to Benzene, which is a compound associated with high cancer risk. The FDA stresses that exposure does not ensure cancerous development, it's a risk factor.

Just like driving without a seatbelt does not ensure accident from injury, or even an accident. It's simply that given other unfortunate factors, the risk impact is significantly heightened.

Food consumer.org reported the following on this matter.

Internal documents from 1990 show that the [American Beverage Association] industry knew that sometimes there levels such as 25 ppb "off the shelf" and 82 ppb after exposure to heat and light. Another popular product, for example, measured at 1 ppb "off the shelf" and 41.5 after exposure to heat and light in a weatherometer.

In a letter on FDA.gov, the FDA does appear to recognize that they have known about such a risk.

Around 1990, FDA was informed by the soft drink industry that benzene, a carcinogen, could form at the ppb level in some beverages that contained benzoate preservatives and ascorbic acid (vitamin C). After learning that benzene was present in some products, research was undertaken by both FDA and industry to understand the factors that contributed to benzene formation. We learned that elevated temperature and light can stimulate benzene formation in the presence of benzoate salts and vitamin C, while sugar and EDTA salts inhibit benzene formation. Contrary to your statement that FDA suppressed information, FDA published its findings in 1993. These findings showed both that benzene was detected only at insignificant levels and that trace levels of benzene could occasionally be detected in foods that did not contain added benzoates and vitamin C.1 Earlier, in 1992, HPB published its findings of a survey in which they sampled fruit, fruit juices, fruit drinks, and soft drinks with and without added benzoate.2 Results of HPB's survey were consistent with our findings.

http://www.cfsan.fda.gov/~dms/benzltr.html