But you may still have trouble detecting GFP this way if you transfected the cells. You can increase your sensitivity to detect it if you stain your cells with an anti-GFP antibody tagged to a different fluorophore in another channel (like texas red or APC).
This also helps when you're using cells that have autofluorescence in the GFP (FITC) channel.
I think Daniel's approach is overkill, and honestly could likely result in false positives. If you have GFP expression, you should see it--this has been done for decades. If you don't, then you have other major issues, such as low transfection efficiency, codon-optimization problems on your protein sequence, or simply unhealthy or dying cells.