Mathematical and statistical methods currently used in improved descriptions of oil reservoirs are surveyed. These methods are used as follows: (1) to discriminate between strata of different properties, (2) to group strata with similar properties, (3) to calculate effective values for simulator input parameters at a simulator grid bloc scale by using core data (scale-up), and (4) to interpolate between wells for predicting values to be used in the interwell grid blocs. For appropriate use of most of the methods described, more data are needed compared with what is usually available from cored wells. Procedures which include use of well logs and/or nearby outcrop data provide some extra data, based on quantitative understanding of the geological features characteristic to a reservoir. 27 refs., 1 fig.