- A regressive form of Slepian modelling was used to develop predictive models for the key variables associated with three quite different experimental data sets. The first data set provided a time series record of the surface elevation for a moderate random seaway. The second data set provided measurements of the wave elevation at the front of a finite draft deep water platform. This data set was significantly more nonlinear since the local wave field was amplified by the presence of the platform. The final data set dealt with the rate of wave run-up and this involved the derivative of the second data set. The results consistently illustrated the need to have an adequate number of events to use as the basis for the regression model. The study presents guidelines for selecting the initial crossing level which is crucial to the Slepian model development. Once the regression model has been established the model allows one to make predictions for other values of level crossing. It was found that the accuracy of those predictions depends on the accuracy of the initial regression process and that reasonable estimates can be obtained. 1995.