What percentage of PhD theses are rejected nowadays?

I'm only personally aware of one student who failed his PhD defense (this is at an R1 US university). After his advisor refused to approve his thesis, he went over his head and got the department chair to schedule the defense anyway. Results were predictable.

On the other hand, "major revisions" are very common, especially, I hear, in the humanities (in engineering, it's far more common to receive token feedback -- if the committee reads the thesis at all! -- than demands for substantial changes).

Outright failing a student during a defense is an extreme embarassment, for the department, for the PhD committee, for the advisor, and of course for the student, so there is every incentive to ensure that a thesis that goes to defense will pass. Moreover, since most theses these days are compilations of previously-published work, it is very easy to tell well in advance if the student is expected to pass.

So if an advisor has doubts about the quality of a student's thesis, he will either ask the student to spend more time improving it, or "suggest" the student start looking for jobs in industry.

Very small, as every failed PhD defence is also a shame for the professor. As a result, the professor will not allow to proceed with defence of the really weak work. And he will listen for other professors that would usually tell in advance they think to vote against.

Hence, most likely, the following will happen:

• If a PhD student just does not work enough, the professor will not allow to continue studies after some time.
• If a PhD student is mad with some own theory or topic that academic community unlikely to accept, the professor will not allow to defend such a work.
• If it is really a bad luck with your topic, the professor will change the topic.
• If the professor has made a strategic mistake and your diligent work does not give results that could be published in a good journal, the professor should normally try to publish anyway in less reputable sources, good enough for PhD defence.

The PhD supervisor is more interested in your success than a lecturer is interested in the progress of the student. Same professor that writes low grades with relatively little attention (as long as he is sure the student deserves) will spend more time when acting as a PhD supervisor, will try to help, will try to fix the topic. This is because PhD project is also his research project. And who would want ones research project to fail? Of course, the professor tries to find a good PhD student for his project, or, if this was not successful, at least to fire lazy or uncooperative student in the first year. But this is way before the actual PhD defence.

If to ask differently, how many PhD students do not get they degree at the end, this really depends a lot on the traditions inside the institution. However in all places I have seen this was below 20 % or about. The first post doctoral position is also seldom a problem.

The next serious threshold you will need to pass is the professor position or at least a permanent researcher position, if you want to stay in science.