floating point exceptions instead of INF, NaN

hi

In my VC++ project floating point division by zero operations (and overflow/underflow etc. ) used to generate 1.#INF000000000000, NaN etc.
Recently for some reason the same operations create an error message:

Unhandled exception at 0x0341f550 in xevi34_2.exe: 0xC000008E: Floating-point division by zero.

Does someone have an idea why this could happen? I am using VCC Studio v8 (2005).

Interesting: If I create a new project and try the division by zero there it still gives back the 1.#INF000000000000 value.

Thanks!
sor73naf