I like to have between .0003 and .0005 actual clearance on the diameter. The thing is, bullet diameter does vary a few tenths.
I saw this in a couple of 6.5 Creedmoor rifles in the cartridge's early days. Total nominal throat clearance is only a half-thou for this design, and this apparently defeated some rifle manufacturers, at least initially.
David Emary (retired from Hornady) wrote a fairly long piece for Lyman's
Long Range Precision Rifle Reloading Handbook on the cartridge's background and development and waxes lyrical on PAT minimisation.
"Limit and control as much as possible Principle Axis Tilt (PAT). ................. It is very simple, force the projectile to enter the rifling as straight and true as possible. In other words don't let it tilt when entering the rifling and being engraved.
............ The best way to design a throat is to provide approximately 0.0005" between bullet and throat."
And so on and so forth.
I soon found that the 139 gn Lapua Scenars I initially tried loading in my first factory Creedmoor, a newly introduced Savage 12 LRP in the chambering, wouldn't pass through the throat without a hard shove as they were an interference fit. Measuring the bullets came up with 0.2645", and the throat was apparently marginally under that. Most of my 6.5 match bullets exceeded 0.264 diameter and suffered this. Nothing that produced the condition would shoot reasonably. Eventually I eventually found a few overlooked boxes of 140gn Nosler Custom Competitions, a low-BC tangent design that I'd relegated to occasional short-distance use and they were the only bullets I owned that would slip through the throat without friction. Finally, the rifle grouped well, in fact even at 1,000.
I went off the tight PAT criterion, and for that matter, the cartridge as a result of what I consider a stupidly small clearance spec. As there are plenty of happy 6.5 Creedmoor factory rifle owners around, I can only assume manufacturers received feedback on this issue and increased clearances a bit.