It’s not actually zero, but it might look like zero if you inspect it using printf or NSLog by using %f.
According to float.h (at least in Mac OS X 10.6.2), FLT_MIN is described as:
/* Minimum normalized positive floating-point number, b**(emin - 1). */
Note the positive in that sentence: FLT_MIN refers to the minimum (normalized) number greater than zero. (There are much smaller non-normalized numbers).
If you want the minimum floating point number (including negative numbers), use -FLT_MAX.