@RF Guy -- Right, and I'm still around!
Once upon a time before OH&S do-gooder dictators came to power, in the days when we climbed transmitting towers to adjust antennas with the RF still switched on, I'd be up there tweaking away, and yet I've lived to tell the tale.
Atop the transmitting tower with my legs wrapped around the mast, the RF arcs would jump from my knees right through two layers of clothes onto the tower. The magenta-violet arcs would burn round holes clean through my jeans and overalls just like the sun and a magnifying glass burn through paper--only much more quickly. The only side effect was some small holes zapped or burned into one's skin around the vicinity of one's knees, and this only occurred because one's clothes prevented direct contact of one's skin with the tower. (Coming into direct contact with the tower prevents any RF arcing from occurring and at VHF/UHF frequencies one simply can't feel electric shocks even though current is flowing.)
My digital watch LCD screen would go completely black with the RF intensity; the only measuring instrument one could use up there was a moving-coil AVO-8 multimeter, as it used a copper oxide rectifier whose frequency response petered out in the upper audio range, thus it remained unaffected by the high intensity RF field.
The radio frequency field strengths that I and others worked in were in the range of hundreds to thousands of volts per metre. Whilst the frequencies were only about one quarter to third that of mobile/cell phones, the radiation intensity was, nevertheless, 10,000 to 100s of thousands of times stronger than that emitted by a single mobile phone.
We were all aware that frequencies with which we worked were all well within the non-ionizing radiation part of the spectrum by MANY, MANY orders of magnitude. Moreover, we too were well aware of the simple heading effects of RF energy and the safety precautions needed. For example, under certain circumstances, RF heading could be dangerous--being accidentally locked in a resonant cavity/tank circuit room with a 100kW or so of RF energy could easily fry one--even though the heating effect was simple molecular (vibrating) agitation.
One precaution we always observed was to take special precautions with our eyes when working with frequencies of 3GHz and above [i.e.: 10cm wavelength and shorter]. At these frequencies (3GHz is the beginning of the microwave band), the wavelength is effectively relatively short. Here, its 1/4 wavelength--the distance where maximum potential difference occurs (where voltage goes from 0 minimum to maximum across the wavefront)--approaches the length of the eyeball. If exposed to excessive RF field strengths at microwave frequencies, localised heating of the eyeball can occur and cataracts of the eye can form.
Well known and normal work practice was never to stand near or in front of waveguides nor to ever look down them, especially so if the RF power level exceeded 1W or above.
Despite this being decades ago, all the people that I worked with in this high level RF field environment are still alive and well.
Radio Frequency energy in the vicinity of mobile phone wavelengths may produce cancer but it's clear the effect is exceedingly small. Over a period of 80 or so years since the 1930s and especially so from WWII onwards--times in which a few people were first exposed to sufficient RF energy to suffer burns--researchers have been looking for evidence that RF energy causes damage to the human body in ways other than straight penetrative heating. Despite hundreds of research efforts all over the world since the war and that there is some theoretical evidence, especially of possibility of harm to young children, it's very doubtful that we'll see a major change in the usage of cell phones.
Frankly, IMHO, the population's addiction to these devices is so total and complete, that it'd take a scare magnitudes larger than some remote possibility of cancer to have any significant affect on usage. Moreover, I don't see power levels coming down much further either, as the same addiction won't tolerate a lessening of the service area, quality and bandwidth throughput.