The number of tests is not what bothers me. The number of times we have had accidents is what does. Thankfully, none has ever resulted in accidental detonation- but this involved a lot of luck.
Eric Schlosser's book Command and Control, which just came out this year is a pretty comprehensive treatment of every accident the US has ever had, from decades of filing Freedom of Information reports. I'm halfway through it right now, it's pretty chilling.
the number of tests should bother you, especially the atmospheric ones. People do epidemiological studies of industrial chemicals and their effects on cancer rates. Nobody does the epidemiological study of how the atmospheric tests affect cancer rates. Because there is no negative control.