Abstarct: Ground control is one of the most effective parameters on production, recovery and safe implementation of room and pillar coal mines. Ground control is the science of studying and controlling the behavior of rock strata in response to mining operation. Generally, the ground control issues in room and pillar coal mines can be divided into two main categories: global stability and local stability. Global stability refers to the stability of a system of pillars. Satisfactory global stability can be achieved by selecting proper pillar sizes according to the given mining conditions. Local stability refers to the stability of openings in the working area including the entry, crosscut, and intersection nearby the pillar being extracted. baxsed on statistics of coal producing countries, the main challenging issue for local stability is roof stability. The evaluation of ground control has always been one of the most important topics for researchers in countries with coal reserves. The aim of this study is to develop models in order to predict and assess the global and local stability in room and pillar coal mines. To achieve this purpose, at first, an intelligent model was developed to predict stable pillar sizing using fuzzy logic theory. This model was constructed baxsed on US room and pillar coal mines databaxse using Mamdani algorithm. The results showed that the fuzzy logic is a suitable and powerful tool for designing pillars in coal mines. In the next stage of this study, two models were developed using fuzzy logic theory (Sugeno inference system) and logistic regression analysis in order to predict the global stability. These models were constructed baxsed on data compiled from room and pillar coal mines, located in West Virginia State, US. It was concluded that both models are suitable and practical techniques that can be used effectively in prediction of global stability with acceptable error rates. Furthermore, the comparison of two models revealed that the fuzzy model was more successful and produced more reliable predictions than the logistic regression model. As mentioned before, the roof stability during retreat mining is the key to ensure the local stability in room and pillar mines. Therefore, in the subsequent stage, a semi-quantitative model was developed to assess the roof fall risk using classical approach of risk assessment. To do this, all effective parameters on roof fall during retreat mining were identified by examination of related literatures. Then, combining all these parameters, a systematic methodology was presented for roof fall risk assessment. In new room and pillar coal mines where mining managers and engineers are unfamiliar with retreat mining and cannot weigh the risk due to the lack of experience, this methodology can be helpful by giving them the capability of using international experience. Ignorance of subjective uncertainties during the process of risk assessment is the most remarkable limitation of the proposed methodology. These uncertainties originate from the linguistic input value of some parameters, low resolution, fixed weighting, and sharp class boundaries. To handle these uncertainties successfully, the fuzzy approach is a versatile and efficient tool. Therefore, in the next stage, this paper presents a novel model to evaluate the roof fall susceptibility during retreat mining using risk assessment fuzzy approach. The proposed model provides a simple and effective mechanism for modeling risk assessment problems involving subjective uncertainties. Finally, to illustrate how the approach works, it has been applied in assessment of roof fall susceptibility in main panel of Tabas Central Mine