Actually, "estimated" is the wrong word to use. The engineering staff calculated a failure rate of around 1 in 100. Management overrode the engineers and guessed the failure rate would be around 1 in 100000. And therein was Feynman's beef about it: the shuttle managers used absolutely no empirical evidence or methodology, they just made sh-- up as they went along. Their risk assessment was non-existent, and their risk analysis was criminal ("Well...the o-ring leaked last time, and the shuttle didn't blow up...so a leaking o-ring's not a problem!")
And THAT particular symptom has gone away. For Challenger, shuttle management was wilfully ignorant - they specifically chose not to understand the technical risks, and ignored the empirical reality right in front of their eyes. Now they have a much better understanding of the technical risks - they just choose to ignore them, or prioritize them as acceptable.
Which, actually, I find perfectly reasonable...as long as the people whose asses are on the line (the astronauts) have the absolute final say. If the shuttle commander says "Yeah, we recognize a risk of dying...we want to go for it anyway" that's one thing. If a project manager in an air-conditioned office in Houston says that, that's entirely different.
My reliability factor is fairly accurate because it's from memory off the top of my head. I could calculate it, using the same methodology NASA engineers do, and it would be dead-on accurate regardless of management stupidity...because it's a technical measure, not a psychological one. The shuttle has about a 96% technical reliability, plus or minus two. When I get back from lunch, I'll put three decimal points on it for you...