Sampling of any geological formation needs comprehensive knowledge of the rock genesis, through the dia- and epigenetic processes (secondary alterations) up to the physical properties and the chemical and mineralogical composition of the rock. Well established theoretical knowledge should be complemented by practical experience. Besides the geological exploration of a raw material, its representative sampling is the most responsible task of the expert. Elaboration of the technology of the mineral processing i.e. the operating cost of the plant highly depends on reliable sampling which is essential for an optimal utilisation of the deposit. Nowadays, when green-field plants below 1 Mt/a capacity are rarely set up, proper sampling is even more important than before. Several million USD/annum can be saved in operating cost with a well-established sampling procedure which, according to the experience of the author, is often neglected by the investors or project management. In this paper the author clarifies the terminology of the characteristic and representative samples and introduces several typical cases both for karst and laterite bauxites.Distinction is made between technologically homogenous and heterogeneous type deposits. Examples are taken from Iranian, Indian, Vietnamese, West African, and South American deposits. Proposals are given for sampling procedures in different stages of the exploration, when the first samples are to be taken for technological tests. The question is investigated whether the bauxite is necessary for an already operating plant (when the bauxite should be fitted to the plant) or when a green- field refinery technology is to be fitted to the deposit(s). In conclusion, it is established that no reasonable economy may exist for mining and alumina processing separately. The economy of the mine and the plant is to be regarded as a single economic unit. For achieving this purpose reliable representative samples are necessary.