rem vs. rad

[Scott Schwahn <schwahn@cebaf.gov>]

I've always read and heard simply that dose equivalent units can only be
used to describe probablistic (stochastic) effects, rather than
deterministic (non-stochastic) effects.  Typically, this would apply
roughly from 0 to about a 0.50 Sv level for a short exposure... 

Inherent in this use must be a linear non-threshold assumption?  But I find
it interesting that with there being new thoughts on a threshold (~0.20 Sv)
for cancer induction (ignoring BEIR VI), that sort of throws the whole idea
of dose equivalent out the window, doesn't it?