In several European countries efforts are undertaken, in particular wi
th regard to fixed industrial installations and transport of dangerous
substances, to quantify the ''societal risk'' (SR) of accidents that
may cause more than one victim at a time. This article explores the na
ture of such efforts. SR-models are essentially ways to structure the
distribution of potential social costs of decisions about hazardous ac
tivities (e.g., costs of risk reduction, of land use forgone). First,
the various ways to describe SR quantitatively, and to set limits to S
R will be presented in short. Next, using a scheme developed by Fischh
off and colleagues, the various approaches will be placed in broad cat
egories of reaching acceptable risk decisions: bootstrapping, formal a
nalysis, and professional judgment. Each of the three categories offer
s a particular appreciation of the risks as 'external costs'. This has
important political implications. In the discussion it is argued that
local SR-limits, by the very nature of SR, should be set in a way tha
t creates consistency with any potential supralocal interests involved
. Second, particular attention is paid to the validity of claims that
SR-limits should reflect a strong risk aversion.