Probabilistic analysis and numerical modelling techniques have been combined to analyse a deep coal mine roadway. Using the Monte Carlo method, a correlation control algorithm and a FLAC 2D finite difference model, a probability distribution of roof displacements has been calculated and compared to a set of measurements from an actual mine roadway. The importance of correlation between input parameters is also considered. The results show that the analysis performs relatively well, but does tend to over predict the magnitude of displacements. Correlation between parameters is shown to be very important, particularly between the three model stresses.