I just purchased a chronograph and am trying to evaluate some new loads I'm working on. What do you consider to be a good value for std. deviation? Do you always get the best groups with the lowest deviation in velocity? I know that seems like an obvious answer, but I have a few loads that seem to work better even thought the velocity deviates quite a bit. Does the deviation go up as you approach or exceed maximum "safe" pressure?

The lower SD load should be more accurate but, it just ain`t so in my experiance.
The low SD might help more at long range where drop from varying velocity could affect your groups but I look more to a low extreem spread as the mark of a mark of a good load. Extreem spread and SD are related but if you watch you can have a low SD with a higher spread.

Elkhunter, I haven't used my chronograph enough to consider myself an authority. I'll tell you what I think I've seen so far.

I shoot a mild load in my 308 - 168 gr. bullet at ~2600 fps. The std. dev'n is about half what I see on my hunting loads. For the hunting loads, I think I see a little more variation as the load approaches the maximum limit, but not much.

I don't see any correlation between group size and std. dev'n. I think the velocity variation is strictly a function of the consistency of your components and your reloading procedure. If std. dev'n has any value, it's as an indicator of your precision in producing the ammunition. For a benchmark, how does it compare with factory ammo?

For my last round of reloading, I weighed and sorted my brass, something I don't normally do. The standard deviation improved significantly. I believe it was because the consistency of the components was better.

As far as I am able to tell, there is no improvement in accuracy by reducing the standard deviation of muzzle speed below about 30 fps.

If you're trying to estimate the long-term muzzle speed SD from a sample of three or five, there will be a very large amount of uncertainty in your estimate, i.e. about plus or minus 50%, IIRC.

Ol' Joe is absolutely correct that extreme spread (range) is closely tied to SD, especially when the samples are small. For five or less, you are practically as well off using range, and, for most people, it is more intuitive. On small samples, SD tends to underestimate variation. Range tends not to.

http://www.shootingsoftware.com/ftp/Perverse%20Nature%20of%20SD.pdf

We all wish it were that simple.

If varying charges to find best harmonics in barrels is necessary, then SD has to take a back seat.

I shot 9 rounds of two different loads. Load #1 shows a max spread of 41 fps and a SD of 15.8. Load two shows an extreme spread of 96 fps and a SD of 31. Both shot about the same.

Does wind velocity affect your chrono readings? I was just thinking that the wind was blowing at about 30 mph (44 fps) from driectly behind me. My chrono was 10 feet in front of the bench. It doesn't seem like the wind would be able to influence the bullet velocity much in the point .003 seconds it takes to get from the rifle to the crony but you never know.

The wind won`t have any effect on your crono readings. Well, maybe it`ll blow the tripod over if strong enough and cause you to lose one.
The 30 mph wind can`t even start to catch the 3000 mph bullet from behind let alone add any push. The wind will cause drift when comeing from the side and can cause the bullet to "slightly" rise or drop from up or down vectors in a headwind, but you`ll see no change on your crony.

A second thought on your SD question.
I saw a thread on SD on another board and a poster offered a way of rateing the SD of a load as either exceptable or not. He used a value he called Coefficient of Variation - the Standard Deviation as a percentage of Average - His thoughts were a CoV of 1% for first class match ammo, and that 2% was pretty good for other purposes.
I`ve no ideas either way about it, but if you like playing with numbers it`s as good a way as any to rate your SD.