I've been thinking a whole lot about Rory's numbers since he posted them, and the importance of going with my read. While my range didn't include the A Jx that he held, it certainly was very very unlikely he had some suited A.

I think you are totally correct about the EV-variance trade-off of taking a card. I did that for flawed mathematical reasons. I didn't want him to be in a big-pot situation where somehow, I gave him odds to get all-in on the flop since it would become obvious I had no flush draw myself. However, this was small-stack thinking. We had enough behind that I could make a big raise, and he'd still not be able to get all of his chips in right there with any sort of good odds. In shorter stacked games, that would probably be a moment where he could move all-in for almost all of his chips and still be getting a pretty good price; that wasn't the case here.

As for showing the hand, in fact, it did help, because he was particularly surprised when he saw my hand and started starting at the stub. I got some real information, that, if I had more clearly thought things through, I think I would have put it all together.

I agree, however, that generally that makes for a bad table image to do those sorts of things. I don't do it often at all anymore (although I used to). I did it here only because I was so on the fence. In the end, it wasn't worth it because even the information I got from showing the hand wasn't enough to push me to the right move.

So, I am ultimately convinced I should have raised on the flop, and be willing to put him on A Ax and back the hand with my whole stack, and just live with it when it turns out he has KKK. My read that he had the A (probably naked) was strong; I lacked the courage of my convictions.

This is probably a good spot to work on my game. I am often a worst-case-scenario player, and give too much credit to opponents when boards look scary, even though previous action makes it unlikely that I could be beat.

Thanks all, for your useful comments and advice.