[Vince Ferrer]

For a complete overview of NASA's probability engineering, see Richard Feynmann's account of his experience with the Challenger commission.

That's a good starting point. I was actually assigned the Challenger hearing once in a class.

Personal observations on the reliability of the Shuttle

The realiability estimates have been known by NASA management to be public relations more than science since Apollo. Apollo program managers basically had reliability requirements imposed on them by politicians. There was a complete disconnet between NASA and congress that you can't claim 1/1000 safety until you test 1000 parts. There would never be the budget to test 1000 of any of the major components of Apollo, individually or as an entire system, and so reliability figures became more estimates and projections to keep the money coming in than based in an engineering reality that other engineers in other industries lived in.

[SunkenCiv]

Thanks VF!