Abstract
The distribution of low-probability, high-consequence occurrences such as industrial pollutant releases according to the Pareto distribution is addressed theoretically. The Pareto distribution represents the maximum entropy distribution for events whose size is proportional to the product of the sizes of earlier events. Additional variables are in operation for events with differing distributions. The Pareto distribution was applied to an actual oil spill distribution. The actual distribution was also produced in a simulation of the effects of 30 arbitrary exponential random variables. The potential effects of the 1995 Komineft oil spill are estimated in the absence of a bypass pipeline using the Pareto distribution of oil spill sizes.