Geo-mechanical parameters of intact rocks such as uniaxial compressive strength (UCS) and young’s modulus (E) are the most essential for characterizing a rock mass, stability analysis of underground and surface structures, and the design of engineering projects. The determination of these parameters is expensive, time consuming task and in need of special laboratory facilities. So, by using nondestructive methods, one can indirectly estimate the cited parameters. In this study, in order to predict geo-mechanical parameters, three methods including multiple linear regression (MLR), artificial neural network (ANN) and support vector machine (SVM) are applied. The data set of different limestone rock types from five different dam sites (Karun 4, Seymareh, Khersan 1, Khersan 3 and Ilam dams) located in Asmari formation, are obtained and analyzed for developing predictive models (with one, two, three and four independent variables). In these models, physical tests (density and porosity), and ultrasonic tests (primary and shear wave velocity) are input parameters. It should be noted that these two tests are primary, cheap, nondestructive and usual tests on rock samples in laboratories. Coefficient of determination (R 2 ) and root mean square error (RMSE) are two criteria of comparison and also selection of optimum method. According to the study results, R 2 values of equations are gradually increased from two independent variables to four unlike RMSE; so, there are very good correlations with four mentioned inputs parameters for estimate of UCS and E (with the respective R 2 = 0.91 and 0.88, RMSE = 4.16 and 3.30). Moreover, SVM can control the error rate and also can perform for high-dimentional data. On the other hand, SVM run time is considerably faster, training is partly simple, and it yields the highest accuracy (largest R 2 and smallest RMSE). To sum up, when the results are taken into consideration, SVM models surpassed the ANN and MLR models.