The relative safety is often brought up in this debate. It's technically correct, but fails to account for just how much damage a single incident can do. Chernobyl had 31 "direct" deaths and several hundred more due to radiation poisoning.
That looks good on paper, but does not account for the radioactive plume that has spread all over Europe. To this day I can find contaminated mushrooms and areas of increased radioactivity in Germany, about 1500km away from the reactor site. 30km around Pripyat are considered uninhabitable for 20000 years and cancer rates in Belarus and Ukraine are still way higher than elsewhere. The Dniepr is contaminated and the flow went all the way to the black sea.
It looks similar in Fukushima: Few direct deaths, a small exclusion zone, but the effects can be measured in the entire Pacific.
We must realize that even if nuclear power works out well w.r.t. Co2 and direct death count, it is fundamentally impossible to rule out the occurance of a catastrophe on par with Chernobyl or Fukushima. If something like that occurred in central Europe, mainland China or one of the US coasts, highly developped regions would be rendered partially uninhabitable and largely undesirable. This kind of damage is not accounted for in these kinds of articles.