NNadir at DailyKos has an analysis of radionuclides and what we really should and should not be worried about them.
To keep the nuclear risks in perspective remember that coal causes over 1 million deaths per year from particulates causing heart disease, lung disease and cancer. Plus 10,000 per year from coal mining deaths. 27,000 of the deaths per year are in the USA which is 25 times the number of US deaths in Iraq.
According to the link on tritium cancer risk, every time a person drinks 1 picocurie of tritium (one trillionth of a curie) his or her risk of cancer increases by 4.4 one hundred trillionths. (See the box on the bottom.) It can be shown that 1 “tritium unit” is the equivalent of 3.2 pCi per liter. Multiplying 3.2 pCi per liter by 5000 by the cancer risk we find that the risk from tritium of drinking a liter of water in 1963 was about 1 in 1.4 billion. World population in 1963 was about 3.2 billion. If we assume that the average person drank 2 liters of water per day each day of the year, it is easy to estimate that the number of cancers induced amounted to about 1600 people.
Today the number of tritium units found is about 20, which corresponds to a risk per liter of around 1 in 360 billion per liter of water. World population is much higher, about 6.6 billion. It follows that the number of people who die this year from tritium, again assuming two liters per day of water, will be a little over 13.