Two methods of Monte-Carlo simulations, one suggested by Matheron (1972) and the other by Rudenno (1983), have been tested for simulating coal deposits. To do this 2160 values of ash percentages were simulated using 60 known data points in a two dimensional grid, 20 m x 40 m. 1L is found that while both methods conserve the mean of the known distribution in the results of the simulations, only the first method reproduces the spatial covariance (i.e. the variogram), of the experimental data. The results show that for simulation of regionalized variables only the first method tested yields the desired results.