A Non-Crappy Gaussian Distribution Calculator

I'll save you grief and time. You want something accurate to 9 decimal places. That's my assumption.

If your range is symmetric, just do this : assume mean is 0, sigma is 1 (yes, normalize before you read further). Now, you want to know how many samples can be outside of -x,+x. How?

Go to to http://www.wolframalpha.com/ and enter erf( x / sqrt(2) )

Of course, having come here, you're smart enough to put your actual number instead of x.

If you want something that can't quite hack 10 digits, just go to

http://calculator.tutorvista.com/normal-distribution-calculator.html

Even the Solido genius Amit can't do better than this.