Polymethyl methacrylate (PMMA) is a useful material for dosimetry phantoms in many ways including approximate tissue equivalence, stability, accessibility and ease of use. However, recent studies indicate that PMMA may have some unanticipated variation in backscatter from one phantom to another. While the reasons behind the variations have not been identified, it has been demonstrated that the backscatter from one phantom to another may vary by as much as 15%, resulting in a dosemeter response variation of as much as 5%. This unexpected contribution to uncertainty in delivered dose to a dosemeter may be quite large compared to the normally estimated uncertainty, potentially causing problems with calibration and performance testing. This paper includes data supporting the differences in backscatter among phantoms, and results from tests on the phantoms performed in an effort to identify possible causes.