Dawn,
In this unit conversion scenario your result is always going to be floating point values, unless all pixels evenly divide by 100 (since you technically could divide by 100 - not a float value).
When you do this operation you're actually decreasing the range of values in the raster. Storing 16 bits may not be necessary. What is the new range of values after multiplying by .01 and doing INT? What format are you working with? ESRI GRIDs are always 32 bit anyway...
Eric