The evaluation of a coal seam for profitable extraction requires the estimation of its thickness and quality characteristics together with the spatial variability of these variables. In many cases the only data available for the estimation are from a limited number of exploration and feasibility drill holes. Spatial variability can be quantified by geostatistical modelling, which provides the basis for estimation (kriging). In cases where the spatial variability of the seam thickness and quality characteristics has a significant impact on how the coal is extracted and stored, geostatistical simulation may be preferable to geostatistical kriging methods. The aim of this paper is to present an improved approach to resource risk assessment by propagating the uncertainty in semi-variogram model parameters into the spatial variability of coal variables. We show that a more realistic assessment of risk is obtained when the uncertainty of semi-variogram model parameters is taken into account. The methodology is illustrated with a coal seam from North-western Spain. (C) 2012 Elsevier B.V. All rights reserved.