The U.S. Bureau of Mines is developing an intelligent system for roof control that uses both an expert system and neural networks to improve the capability of mining engineers to evaluate roof support effectiveness for ground control in coal mines. The expert system compares roof support capacities with the support requirements estimated to be necessary to maintain entry stability. It does this by evaluating the results of tests on various types of roof support and establishing the maximum allowable load according to anchorage capacity and yield strength of the support. After the user enters geological information (rock properties, geometry of the opening, in situ stresses, bolt pattern parameters, etc.), the expert system compares the predicted required loading to the roof support capacity and gives the operator advice on the adequacy of the design and how improvements could be made. A good source of real-time data necessary to allow the expert system to make decisions will come from a roof bolting machine being developed by the Bureau of Mines. Using this information, two neural networks were developed to identify different types of strata and features in a mine roof, such as rock type, rock compressive strength, and joint characteristics.