Answers to many questions arising in the different phases of exploration, estimation and production of a mineral deposit, can be found in the geostatistical theory of G. Matheron. It is shown that all these questions regarding sample size, number of samples, shape of drilling pattern can be answered as soon as the continuity of the mineralization is quantitatively characterized by means of a variogram. Next a formula which gives the variance of the error associated to any estimation procedure (i.e., any method and any sampling pattern) is given. Applications to the computation of the standard error of the'mean for various sampling patterns are shown, as well as how to use the formula to simulate several possible sampling strategies and choose the best one. Altogether it is shown how to answer the often heard question, "what happens if such and such sampling procedure is used?", rather than giving an hypothetical "optimal" solution.