In this study results of an investigation into ground vibrations of an underground excavation in southwest Iran has been discussed. Recorded experimental blast data have been analyzed employing two different methods of analysis. A comparison between the two ways of investigation, namely empirical equations and neural network, is presented. It has been shown that the applicability of neural network method is, by far, more promising than any of three selected empirical equations. It was also found that, in spite of releasing high correlation of determination (R2), empirical equations face discrepancies with real data in high range of vibration intensity whereas neural network fit the data of all ranges with a consistent accuracy.