Discriminant analysis is an application of multivariate analysis, which may have its use in determining accident risk levels and premiums of industrial enterprises. This paper only aims to give some suggestions. The following questions will be considered.

1. Determining a discriminant function which makes it possible to discriminate between the risk levels of the industrial branches in an efficient way. The industrial branches comprise enterprises with comparable risk levels, hence they are to be considered as homogeneous groups. The function will at the same time serve as a means to classify seperate enterprises into one of these groups.

2. Fixing risk functions which enable us to rank the enterprises of an industrial branch to increasing risk on the ground of observations of a number of variates which characterize the risk situation.

3. Using these risk functions to calculate premiums. The classification-question mentioned under I was the reason to consider the technique of the discriminant analysis. By virtue of the Dutch Industrial Accidents Act every five years a tariffdecree is being published. This decree contains the premiums per wage-unit for the industrial branches. However, there are enterprises e.g. large compound enterprises which do not fall under these regulations. These enterprises ought to be classified according to their own experience. That means we need the knowledge of the risk levels of these particular enterprises in relation to the fixed risk levels of the industrial branches. As mentioned, this is a problem inherent to the typical Dutch situation. It seems, however, probable that such problems and the techniques we intend to sketch have a wider and more general meaning for the accident insurance.