The issue is how the cable is manufactured into the plug. I think the plug is good for a big voltage, but there will be some form of discontinuity in the dielectric, and it can spark across that if it is an air gap and is not filled somehow, or continuous into the plug.
FWIW: I use F-type satellite connectors and high quality German 4-screened (2 solid, 2 braided) satellite coax. I put an account up on fusor.net. I modify a female-female coupling so that the cable dielectric fits snugly into it. These days, I am regularly running these to 15kV and have never had any indication of a breakdown.
http://www.fusor.net/board/view.php?bn= ... 1258401294(I've not shown refitting a shell, but once it is screwed up it just looks like any other F-connection.)
It is my standard coupling now. Because I have a need for so many RF couplings in my system, it was the only thing to do to avoid big cost. (Besides, you can't get 15kV BNC!) I've struggled to work out a means for a suitable, robust 'conversion' back to BNC at those voltages, thought, so I just don't use BNC's for any high voltage work at all.
When I did testing on these, I ran totally unmodified connectors - just to see what the 'minimum' was, and they still took 3.3kV before breaking down (in the air gap between the central wire and the coupling - which this mod fills).