Evaluating the safety record of something which due to extreme safety measures fails only very occasionally, but renders cities unsafe for human habitation and livestock on the other side of Europe unfit for human consumption decades later, clearly requires a different approach to evaluating risk than something which kills people only in regular small industrial accidents during manufacturing and on-site erections.
The known safety record of nuclear power thus far is pretty good, which is mainly because of huge investment in safety driven by paranoia. But even looking purely at the catastrophic failure scenario as the only nuclear-related cause of death, if you design nuclear power stations to be able to withstand natural disasters that only occur once every hundred years then you might have to wait sixty years for the first breach and another sixty for the first bad unavoidable breach. Making a call about the long term safety of something which can potentially cut short the life of millions of people just because it hasn't happened yet[1] could be a little premature, although with any luck we'll have something sufficiently better and safer than U235 fission to render most of today's power generation technology obsolete in a couple of decades time anyway.
[1]If you take the most extravagant claimed death figures published by scientists ["Chernobyl: Consequences of the Catastrophe for People and the Environment" - Yablokov; V & A Nesterenko], Chernobyl could have been responsible for the premature deaths of as many as a million.
Of course, as things age, problems become far more likely.
It's stupidly irrational how people require hundreds and hundreds of time more proof for a power source that has OMGRadiation!!1! but require no evidence of the safety of other much much more dangerous sources of power.