Abstract
This paper reviews current practice with respect to identifying hazardous locations and assessing accident countermeasure effectiveness and, in particular, the use of the poisson distribution. An analysis of accident count data is described. The result revealing grounds for doubting the general validity of the widely- used "poisson assumption".The paper describes alternatives to the "poisson assumption", so that when analysing temporal variations in accident occurrence, one can take account of the variations in the variability of accident counts.