I've been trying to get a handle on radiation exposure levels. As LiamC points out, the type of radiation makes a big difference, and most importantly, whether it's on your outside or inside. Ingestion of alpha particle sources is highly dangerous because the radiation will all be absorbed by your tissues. Most of the Chernobyl thyroid cancer cases stemmed from drinking milk produced by cows that ate contaminated grass ...
As a completely unscientific exercise, I have come up with a quick and dirty rule of thumb that tries to relate whole of body exposure to an increased risk of dying from cancer. It's based on what's known statistically about the effects of medical radiography, but please don't take it too seriously.
Increased risk percentage = millisievert dosage divided by 100.
Firstly, you have to realize that your chances of dying from cancer are about 25% anyway. If you somehow manage to be exposed to the recommended maximum 1 millisievert dose every year for 70 years, that equates to a minimal additional risk of 70/100 = 0.7%. So nothing to see here.
If you're a nuclear industry worker (eg uranium miner), the maximum recommended level is apparently 50 millisieverts per year, although normal exposure is expected to be much lower. If you rack up 20 years service in a hypothetical shoddy workplace where you always experience that maximum dose, your risk of fatal cancer moves from 25 to 35%. In practise, I'll bet exposure would more likely average 10 millisieverts, so that's only a small 2% extra risk.
Chest x-rays these days use really, really low amounts of radiation (I've read 60 microsieverts, i.e. 0.06 millisieverts), so you don't need to worry about them at all. However, CT scans are an entirely different beast; thorax scans may be between 2 and 8 millisieverts and a barium enema 15 millisieverts. You wouldn't want to have too many of these procedures without good cause. My rule of thumb overstates the risk by about 2.5 times at this point (we don't live in a linear world). Adjusting for that, one out of every 667 people who receive a barium enema will theoretically die from a cancer caused by the scan. Of course, they're far more likely to die from whatever illness they had that required the scan in the first place.
At the extreme end, a 10 sievert (10,000 millisievert) exposure works out to be 100% fatal. 5 sieverts gives you a 50% chance (about 30 of which is death from radiation sickness, but hey, I'm trying to keep it simple). One sievert is 10%, so that's starting to approach Russian Roulette odds.
Over the last couple of days, measurements at Fukushima Daiichi have been reported as 400, 1000, and 600-800 millisieverts per hour. So using my ready-reckoner, a worker who is exposed to the full 400 for 6.25 hours just doubled his chances of fatal cancer. 100 minutes at 1000 millisieverts can be compared with spinning a six-chambered revolver with one bullet.
I guess that would be why they evacuated the workers earlier today.