Several empirical equations for estimation of coal pillar strength are presented in academic studies so far. The development processes of these equations are almost similar and are usually obtained by fitting the mathematical function (curve) on field data. One of the best criteria to evaluate the quality of fitting for such equations is the correlation coefficient, R2, which nevertheless has limited applicability. For calculating the correlation coefficient access to the initial data that the equation is presented by fitting on them is necessary; for many coal pillar strength formula it is impossible to conduct this. In this paper a new approach based on the analysis of uncertainty amplitude to compare the coal pillar strength formula are presented. This approach is based on a combination of parameters such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), type of function and degree of freedom. Subsequently the confidence level of constants is formed and thus the correlation coefficient becomes more comprehensive. Therefore for an effective comparison the efficiency and accuracy of coal pillar strength formula can be used.