It might help to know what is fluorescent in your assay. Does it have ionizable groups? The pH could be having a big effect. With just water, the pH is uncontrolled.

Were the PBS and HEPES buffers made the same day the experiment was conducted and were the buffers sterile filtered?  I've found doing this fixes 90% of fluorescence polarization assay issues.  Also, what are the excitation and emission wavelengths?

I am performing FP on 3'-FAM labeled DNA and RNA aptamers which were generated against a ~100 kDa soluble protein.  The PBS was a little old but the HEPES buffer supplemented with MgCl2 was made the same day as the experiment (these were not sterile filtered however). The excitation and emission wavelengths are 488 and 525, respectively.  Thanks a lot I really appreciate the information.

FAM fluorescence responds to pH, losing fluorescence at acidic pH. Buffering your experiment at pH 7 or above will be important. In addition to pH, the higher ionic strength of the buffer as compared with water could affect the structure of your aptamer, formation of intermolecular aptamer aggregates, the binding affinity between the aptamer and its ligand, and the interaction between the FAM label and the oligonucleotide.

Hi,

pH can definitely affect the quantum yield and shift the absorbtion spectra of fluorescent dyes and proteins.

http://www.sciencedirect.com/science/article/pii/0022231375900034 (i see your dye is pretty similar to fluorescein)

http://www.sciencedirect.com/science/article/pii/S0006349598778701 (here they discuss the case of GFP)

so I agree with Adam, try to buffer your experiments to work on more solid ground.

Best,

Jacopo