The link rayban posted is a good one. But I do have one discrepancy with it: the author advises the tang shoulders and the guard slot be square. This is likely the easiest way to get a good fit, but it's also the weakest way to build a knife.
Sharp inside angles, deep scratches, and the like are called stress risers. I've seen many rotary dies fail; I'm talking about 100# or more dies machined from solid tool steel, that rotate in bearing blocks on an integral shaft of about 1-1/2" diameter.
In nearly every instance* that the shaft broke, it had been made with a square transition between the shaft and the body of the die, and that's where it snapped. I've
never seen a die with properly radiused transitions fail this way -- same type of tool, same type of steel and heat-treat, used in the same machines. The only difference was the stress risers.
In short, if the tang shoulders are radiused, that is, a curve instead of a sharp right angle, the likelihood of breakage is greatly reduced. Given reasonably good HT, I'd dare say it's virtually eliminated. (Although of course, you can break any knife, with a solid vise, a cheater bar, and two or three large friends
)
Fitting the guard slot to a radiused tang takes more work and care, but pays big dividends in the strength of the blade, and the strength of the guard/tang joint.
* I did see a couple of similar tools fail when the shaft broke; in those instances they broke very unevenly with jagged edges and a coarse feel to the surface of the break. I believe this has more to with improper heat-treat than anything else, and was very rare.
