Research shows. Studies have found. Scientists conclude. Each of those phrases can be completed, accurately, with any one of the following: That it's possible to use ESP to see the location of someone far away. That exposure to lead at everyday levels does no harm to the developing brain. That hormone replacement protects women against heart disease. My point is not that science is always tentative and that scientists are fallible, though both are certainly true (since all three of the above are wrong), but that almost anyone with an agenda can find research to support it.

Keep that in mind this summer when the Food and Drug Administration issues its report on bisphenol A, the chemical building block of polycarbonate, the hard plastic used for some baby bottles and water bottles, and of the epoxy resins that line food cans. In her first appearance before Congress as FDA commissioner, Margaret Hamburg told a House panel this month that the agency is reconsidering its decades-old position—reiterated last year—that BPA at current levels of exposure poses no harm to human health. You can be sure that if FDA deviates from that conclusion, the plastics industry will deploy the three phrases above, completing them with "—that BPA is perfectly safe."

Whether that's true can be answered only by empirical data. But not all empirical data are created equal. BPA studies that a spokesman for the American Chemistry Council described as "definitive," for instance, have come in for criticism on three fundamental grounds, not including that they were partly funded by industry (I don't reflexively assume that industry-sponsored research is suspect; whether a study is good or not depends on how it was conducted). First, research in 2002 used a strain of rat that is extremely insensitive to estrogen; it doesn't even show hormonal effects if it's given 100 times the dose of estrogen in human birth-control pills. Since BPA acts like an estrogen, finding no effect in this insensitive rat is about as illuminating as not finding an effect of rain on a waterproof watch. That doesn't tell you that water can't harm machinery. Second, a 2008 study found that prostates in mice not exposed to BPA—these were the control animals—were 70 percent larger than normal. That's a problem: other studies have shown that BPA enlarges the prostate by about 35 percent. If you're looking for a prostate effect by comparing BPA-exposed mice to mice with mysteriously abnormal prostates, it's no wonder BPA gets exonerated. Finally, another 2008 study compared BPA to estradiol, a form of estrogen. But estradiol had never been used to provide such a baseline, so concluding that BPA is less potent than estradiol—as industry does—is like saying one temperature is higher than another when you don't even know if the thermometer works.

Evidence on the other side is both stronger and more convincing. I can regale you until I'm out of space with studies showing that in monkeys, levels of BPA at the upper end of what the U.S. government calls safe harm synapses responsible for learning and memory; that people with the highest levels of BPA are most likely to have type 2 diabetes or heart disease; that BPA given to pregnant lab animals permanently alters the expression of genes responsible for uterine development and damages the reproductive system of their fetuses. More telling than individual studies is the weight and quality of the cumulative evidence. Based on that, the FDA's Scientific Advisory Board last year rebuked the agency for failing to consider all credible evidence when it called BPA safe for use in food containers, and the Endocrine Society issued its first "scientific statement," concluding this month that for chemicals like BPA, "the concern is real."

That concern is likely to rise as the FDA takes account of new data showing that people are exposed to more BPA than it assumed when it concluded that exposure to BPA is within the margin of safety. In a study presented at the Endocrine Society, scientists led by biologist Fred vom Saal of the University of Missouri found that monkeys fed 400 times the amount of BPA that the government assumes people ingest had lower levels in their blood than the average American. For BPA levels in people to be higher than in monkeys that practically gorged on the stuff, we must be ingesting way more than the FDA thinks. Where are the higher amounts coming from? In addition to hard plastic and epoxy can linings, it turns out, newspaper ink and carbonless copy paper—the stuff of credit-card receipts and all sorts of business and medical documents—contain high amounts of BPA. Recycled, they wind up in food containers such as pizza boxes, along with the BPA. It's never a good thing if people are exposed to more of a chemical than safety agencies thought, and if studies giving that chemical a clean bill of health are so troubled. As common sense (never mind research) shows.

Begley is NEWSWEEK’s Science Editor.

© 2009