Grade-tonnage curves are important decision making tools at both the planning and operating stages of a mining project. In this study, uncertainty resulting from estimation of these curves is studied by simulation. In particular, the effect of short-scale variability, change of support and type of data distribution on the uncertainty are examined. The ratio R = Col(Co+ C) is considered as a measure of the short-scale variability and simulations are carried out for two extreme values of this ratio. Point support and block support with size of 20 x 20 m(2) are taken into account. Normal and log-normal data distributions are used. As a simulation technique, unconditional sequential Gaussian simulation is considered. The results show that short-scale variability, change of support and type of distribution affect the uncertainty considerably.