Recent studies have demonstrated that the tensile strength of rocks can effectively be used as a design parameter in some subjects such as production method and excavation equipment selection, design of blasting geometry, slope stability, and dimensioning of underground openings. Different testing methods are used for determining the tensile strength of rocks. These methods can be divided into two groups that are direct and indirect methods. According to test results reported by different researchers, the tensile strength of rocks is varying in relation to testing methods. By conducting direct and indirect tests, experimental coefficients which will estimate direct tensile strength from indirect tensile testing methods such as Brazilian, 3-4 point bending tests, are presented in this study. Besides, a geometric approach has been proposed to explain the reason for this difference. This approach claims that it is more important the average stress per unit volume in the tensile zone of the specimen than the maximum stress in the specimen to develop damage. Average stress in the specimen has been expressed as stress per unit volume because each specimen has a different size in different testing methods. To calculate the average stress in a simply way, a geometric shape, which is related to the stress distribution of the specimen and called "characteristic unit volume" has been defined. The volume of the geometric shape has been defined as the theoretical coefficient of the related testing method. For carbonate and igneous rocks, the experimental coefficient was 0.47/0.55 for the Brazilian test, 0.29/0.37 for the 3 point bending test, 0.37/0.43 for the 4 point bending test, respectively. The theoretical coefficient was found to be 0.5 for the Brazilian test, 0.33 for the 3 point bending test, and 0.390.42 for the 4 point bending test. Experimental and theoretical coefficients obtained in this study are in accordance with each other.