Is there any science to what twist rate is correct or is the best you can do is make an educated guess? I currently shoot a lot of 338 cal and a 1in 12 works great for 250s. I know in most calibers a 12 twist is way too slow. I am getting ready to build a gun in .243 cal and it looks like most guns in this caliber are either 8 or 9 twist. Does the required twist rate get faster when you go down in bullet diameter? I do not plan shooting over 80 or 90 grain bullets in this gun.

The short answer is that you can think of twist rates in terms of 'calibers per turn', and some of the mystery goes away.

Most heavy high BC bullets require about 32 calibers per turn to stabilize.

In .22 cal, this equates to a twist rate of 1:7.2"
In .244 cal, this equates to a twist rate of 1:7.8"
In .257 cal, this equates to a twist rate of 1:8.2"
In .264 cal, this equates to a twist rate of 1:8.5"
In .277 cal, this equates to a twist rate of 1:8.9"
In .284 cal, this equates to a twist rate of 1:9.1"
In .308 cal, this equates to a twist rate of 1:9.9"
In .338 cal, this equates to a twist rate of 1:10.8"

So as you can see, the twist rate in inches per turn gets slower as the caliber is larger, but the physics of scale dictate that the bullet needs a given 'caliber per turn' which applies universally to all calibers.

Note the .257 and .277 calibers require twist rates that are much faster than commonly available in those calibers. As a result, those calibers are limited to shooting bullets that are not as proportionally long and heavy as those in other calibers.

The link above goes into greater detail about stability near the end of the article.

WV- I Shoot a rem 700 .243 with a 1-12 twist ( happens to be a hart). lapua brass and berger 88 grain bullets. it shoots unbelievably. most times in the ones. i have not tried the 95's yet. i am totally confident it will shoot 90 baltips though. the twist rate , bullet length and velocity determine stability. barrel quality determines accuracy. you don't say what kind of barrel you are putting on it. a 10 twist will let you shoot up to 100 grainers and will still shoot the lighter ones. my experience. roninflag

Here is something I have saved, dunno who wrote it:

Calculating Bullet RPM from MV and Twist Rate
The lesson here is that you want to use the optimal RPM for each bullet type. So how do you calculate that? Bullet RPM is a function of two factors, barrel twist rate and velocity through the bore. With a given rifling twist rate, the quicker the bullet passes through the rifling, the faster it will be spinning when it leaves the muzzle. To a certain extent, then, if you speed up the bullet, you can use a slower twist rate, and still end up with enough RPM to stabilize the bullet. But you have to know how to calculate RPM so you can maintain sufficient revs.
Bullet RPM Formula
Here is a simple formula for calculating bullet RPM:
MV x (12/twist rate in inches) x 60 = Bullet RPM
Quick Version: MV X 720/Twist Rate = RPM
Example One: In a 1:12″ twist barrel the bullet will make one complete revolution for every 12″ (or 1 foot) it travels through the bore. This makes the RPM calculation very easy. With a velocity of 3000 feet per second (FPS), in a 1:12″ twist barrel, the bullet will spin 3000 revolutions per SECOND (because it is traveling exactly one foot, and thereby making one complete revolution, in 1/3000 of a second). To convert to RPM, simply multiply by 60 since there are 60 seconds in a minute. Thus, at 3000 FPS, a bullet will be spinning at 3000 x 60, or 180,000 RPM, when it leaves the barrel.
Example Two: What about a faster twist rate, say a 1:8″ twist? We know the bullet will be spinning faster than in Example One, but how much faster? Using the formula, this is simple to calculate. Assuming the same MV of 3000 FPS, the bullet makes 12/8 or 1.5 revolutions for each 12″ or one foot it travels in the bore. Accordingly, the RPM is 3000 x (12/8) x 60, or 270,000 RPM.
Implications for Gun Builders and Reloaders
Calculating the RPM based on twist rate and MV gives us some very important information. Number one, we can tailor the load to decrease velocity just enough to avoid jacket failure and bullet blow-up at excessive RPMs. Number two, it helps us calculate the minimum twist rate needed to stabilize a particular bullet. Although there are other important factors to consider, if you speed up the bullet (i.e. increase MV), you MAY be able to run a slower twist-rate barrel, so long as you maintain the requisite RPM for stabilization. (Note: this is a general principle, but other factors, including bullet shapes, Center of Gravity (CG) and jacket thickness can affect stability). The slower twist-rate barrel may, potentially, be more accurate.
That said, we note that bullet-makers provide a recommended twist rate for their bullets. This is the “safe bet” to achieve stabilization with that bullet, and it may also

if you speed up the bullet, you can use a slower twist rate, and still end up with enough RPM to stabilize the bullet

This is a common fallacy with stability. It's not true, and so fails many tests.

You might recognize that a twist requirement is in displacement per turn(12:1).
The displacement is typically assumed standard sea level density air.
There is no time, or velocity rates in this requirement.

Why is that?
Afterall, if technically 'just below' Sg of 1.0(barely stable, possibly), and you bump velocity way way up, you could technically raise Sg to >1.0, and not tumble. Well, because another 10deg drop in temp can again result in tumbling no matter adjusted velocity.
That's one of many tests that could fail here, and the truth always passes all tests.
Also, 'Just below' or 'Just barely' stable performs so badly that it is NEVER considered viable in twist requirements (ideal is Sg 1.5).

So you will never see bullet makers assigning RPM requirements, because it would be woefully incorrect to do so, and cause hardship to many customers.
Sorry for the rant, but this RPM perspective really must be quashed.

[QUOTE=Mikecr This is a common fallacy with stability. It's not true, and so fails many tests.

I agree with Mike on this one. was talking with Gerard Schultz of GS Custom bullets about this and he said the twist requirement barely changes with the speed because the faster it goes, the more RPM's are required.

Bryan Litz(above) explains RPM relationships very well in his book.

My perspective on twist requirements center on displacement as the dominant overturning cause. It is also not the truth in that it will also fail tests. But it passes way more tests than RPM rules of thumb, given a set bullet design(it's Cg to predicted Cp arm).
By displacement, I'm really talking drag throughout it.
And my perspective here is a way of mitigating errors in twist choices.

I see a turn of a bullet(it's inertia) as overcoming x.xx" of displacement(overturning cause). Increasing velocity really doesn't change this model. A 12:1 twist barrel causes 1 turn per 12" at the muzzle, regardless of velocity. And a bullet going down range(slowing) displaces fewer and fewer inches per turn to overcome, allowing it's gyroscopic stabilty to climb continually. The twist rate can go from 12:1 at the muzzle to a relative 6:1 by 1kyds. Sg might end up 3-5.0 by then..
This holds for dynamically stable bullets still above transonic.
As air density goes up,(or drag goes up), the relative displacement goes up, and 12:1 actual turns into a relative 13 or 14:1 in free flight (in my model).

In reality, bullet inertia must overcome other moments, and it can also actually be a source of moments. Muzzle release is a big one, hitting your target is another big one,, traversing transonic, imbalances and flaws, etc.. Alot of things going on that could combine to overtake displacement as dominant, but that's not common these days(much less consistant).

I think the intent of recommended twist rates should be taken in this context, even though there is more to it.