If you are using a pre-XP version of Excel, NORMSINV will produce
totally unacceptable results, but you could use inv_normal from Ian
Smith's VBA code for probability functions
http://members.aol.com/iandjmsmith/examples.xls
XP's NORMSINV does not produce random numbers millions of standard
deviations from the mean, but is still crude enough that it is only
acceptable for casual work. If you don't have 2003 you may get better
results from the Box-Muller method
=SQRT(-2*LN(RAND()))*COS(2*PI()*RAND())
If you have 2003, either method will probably work reasonably well
provided that you have downloaded the RAND patch from Microsoft.
Jerry