The impulse response of models of subsurface formations are statistically analyzed for abstracting seismic parameters which could be characteristic of the stratigraphy and lithology of the formations. Two types of formations have been considered, consisting either of sand-shale sequences or of coal-shale sequences. Models of the formations are generated using the Monte Carlo method. It is found that three features in the power spectrum of the impulse response, namely the frequency fe at which the spectrum can be divided into a zone of high energy from a zone of low energy, the lowest frequency ,fp where there is a significant energy peak, and the frequencyf, at which there is a maximum energy, can be used statistically to distinguish between the formations consisting of sand-shale sequences and the formations made up of coal-shale sequences. Three additional parameters AZ/AI, AZ/AC,, and AI/A o, where A denotes the autocorrelation function of the impulse response and the subscripts denote the lags, are also statistically significant discriminators between the sand-shale formations and the coal-shale formations. The discrimination between the two subgroups of each model consisting of more (or less) than fifty percent of one lithology is also feasible, although there are fewer discriminants available.