The day before Thanksgiving, the busiest flying day of the year, was chosen as "National Opt-Out Day" to encourage people to ask for a pat-down from the TSA rather than go through the new Backscatter Scanners. The organizers of Opt Out Day seem primarily concerned with privacy--people in the backscatter photos look kind of like creepy naked mannequins of themselves--but the scanners' safety has been questioned as well. I've done a lot of X-ray work as a materials scientist, so I'd like to summarize what we know.
This enticingly titled Ars Technica article does a great job of describing the difference between conventional medical X-rays and the new backscatter detectors, but ultimately concludes that the "biology" part of "physics and biology" is not well understood. What X-rays and cosmic rays and ultraviolet radiation from the sun have in common is that they are ionizing radiation, electromagnetic radiation with high enough energy to knock an electron off of an atom. Visible light, wi-fi signals, and microwaves are also electromagnetic radiation, but are not energetic enough to ionize atoms. Ionized atoms are more chemically reactive than neutral atoms. If an atom is ionized in the living cells of your body, the atom may then react with nearby atoms in such a way as to render a useful molecule defunct, or worse, create a mutation in a DNA molecule.
Much of our information about the health effects of ionizing radiation comes from Hiroshima and Nagasaki. It's grim to think of, but the conditions were perfect: hundreds of thousands of people were affected, and they could precisely report their distance from the explosions. The units and terminology are confusing: the Curie is a measure of radiation itself, essentially the number of photons, but it's almost never used. Instead, units like the rem (Roentgen Equivalent in Man) and the older rad (Radiation Absorbed Dose) include a correction factor to account for the likelihood that a specific energy of photon will be absorbed.
In my radiological safety training, I was told that the information from Hiroshima covered doses ranging from "very high" to "massive". Lower-dose cases were obsured by the citizens' ongoing exposure to radioactive dust and debris. Medical researchers extrapolated health effects downwards from "very high" doses to the smaller doses one might encounter in dental X-rays and industrial accidents. Unfortunately, the accuracy of that extrapolation was hard to verify, because there were fewer people involved and the circumstances were less controlled. Everyday doses from sources like radon are orders of magnitude lower, and their health effects correspondingly less clear.
It's comforting to know that, according to the Ars Technica article, the backscatter scanners give a dose of only three microrem, about a thousandth of the dose you'll get from cosmic rays if your flight crosses the continental US. But then you realize that this dose is concentrated in a small volume: your skin. You could try to do the math and say that three microrem in the outermost few millimeters of your body is equivalent to some larger dose distributed evenly throughout the body. The truth is, the health consequences of radiation aren't well enough known to make that kind of calculation - or the assumption that risks scale linearly down to low doses in the first place.
We're exposed to radon, cosmic rays, and radioisotopes continuously throughout our lives. Somehow we manage to have kids - and, occasionally, tumors. Do biological cells have a certain rate at which they can repair radiation damage without permanent harm? A nonlinearity in the risk curve? If so, we don't know what it is, partly because we engage in so many other activities that are riskier.