A definition of the term 'Computational Dosimetry' is presented that i
s interpreted as the sub-discipline of computational physics which is
devoted to radiation metrology. It is shown that computational dosimet
ry is more than a mere collection of computational methods. Computatio
nal simulations directed at basic understanding and modelling are impo
rtant tools provided by computational dosimetry, while another very im
portant application is the support that it can give to the design, opt
imisation and analysis of experiments. However, the primary task of co
mputational dosimetry is to reduce the variance in the determination o
f absorbed dose (and its related quantities), for example in the disci
plines of radiological protection and radiation therapy. In this paper
emphasis is given to the discussion of potential pitfalls in the appl
ications of computational dosimetry and recommendations are given for
their avoidance. The need for comparison of calculated and experimenta
l data whenever possible is strongly stressed.