The competitive coal market has required coal producers to put more effort into coal quality control. Successful quality control requires a good estimation of the in situ coal quality as well as a good estimation of the probable fluctuation of the in situ coal quality during production. Geostatistics offers a very promising way to meet this requirement. A geostatistical approach toward the coal quality control problem is presented. Two models, the prediction model and the conditional simulation model, have been developed for coal quality data from two large export coal mines in central Queensland, Australia. The mine planning and scheduling aspects based on this and other potential applications of the models are addressed.