Recently the compensator has been shown to be an inexpensive and relia
ble dose delivery device for photon beam intensity-modulated radiation
therapy (IMRT). The goal of IMRT compensator design is to produce an
optimized primary fluence profile at the patient's surface obtained fr
om the optimization procedure. In this paper some of the problems asso
ciated with IMRT compensator design, specifically the beam perturbatio
ns caused by the compensator, are discussed. A simple formula is deriv
ed to calculate the optimal compensator thickness profile from an opti
mized primary fluence profile. The change of characteristics of a 6 MV
beam caused by the introduction of cerrobend compensators in the beam
is investigated using OMEGA Monte Carlo codes. It is found that the c
ompensator significantly changes the energy spectrum and the mean ener
gy of the primary photons at the patient's surface. However, beam hard
ening does not have as significant an effect on the percent depth dose
as it does on the energy spectrum. We conclude that in most situation
s the beam hardening effect can be ignored during compensator design a
nd dose calculation. The influence of the compensator on the contamina
nt electron buildup dose is found to be small and independent of the c
ompensator thickness of interest. Therefore, it can be ignored in the
compensator design and included as a correction into the final dose di
stribution. The scattered photons from the compensator are found to ha
ve no effect on the surface dose. These photons produce a uniform low
fluence distribution at the patient's surface, which is independent of
compensator shape. This is also true for very large fields and extrem
ely asymmetric and nonuniform compensator thickness profiles. Compared
to the primary photons, the scattered photons have much larger angula
r spread and similar energy spectrum at the patient's surface. These c
haracteristics allow the compensator thickness profile and the dose di
stribution to be calculated from the optimized fluence profile of prim
ary photons, without considering the scattered photons. (C) 1998 Ameri
can Association of Physicists in Medicine.