Abstarct: Several researchers tried to investigate rock joint roughness by simplification and applying related assumptions. In Barton and Laubscher’s presented methods the rock joint has to be compared visually with typical models which related to predefined number of roughness. Several researchers also used statistical methods considering quantitative estimations. Visual methods are simple and user friendly but they are subjected to errors which is related to judge in achieving best fits and distinction the amount of roughness of one joint to another. In quantitative methods applying one quantitative parameter can’t accurately express the all specification of joint roughness. In this research two methods have been proposed in order to determine joint roughness coefficient by applying the shape of joint. In first method, by gathering rock joints surface data include 203 samples from site investigations and literatures its tried to define all probable geometry of rock joints. The shapes of probable geometries have been studied and shapes with more frequency introduced in 20 geometry model named representative geometry model (RGM). The representative geometry models have been evaluated and categorized with quantitative parameters such as Rq, Rp, Z2 and SF. In this category, by comparing the rock joint surface with RGM’s and considering parameter K the more accurate estimation of Joint roughness coefficient can be achieved. In second method, by statistical study of the condition of rock joints surface in 10 centimeter length a new category proposed for rock joint roughness. The surface of rock joint divided in four parts and for each part four level of height considered and accordingly 256 probable cases of roughness surface have been achieved. These probable cases presented in a new category named Abacus category that cover all probable kinds of rock joints. In this method, several groups of the presented category have been categorized by considering of maximum height of surface, differences of adjacent heights and frequency of adjacent height differences. The presented groups in this category by using an appropriate data (112 samples) of roughness joints which gathered from literatures have been attributed to joint roughness coefficient (JRC). The category also has been developed for six columns and results have been presented. In this model, Equations display higher correlation coefficients (0.99). By applying the category which proposed in this research, the amount of JRC can be achieved in a simple way, short time and with appropriate accuracy.