In this work, results are presented on the application of standard LIBS and Nanoparticle-Enhanced LIBS (NELIBS) to the classification of rocks (igneous and sedimentary). The classification of the spectra obtained with the two methods was performed using Principal Component Analysis (PCA) and Graph Theory method. The results obtained confirmed the advantages of the LIBS technique in geological applications, showing that excellent classification of the rocks analyzed (more than 99% of the spectra correctly classified) can be obtained using standard LIBS coupled to Graph Theory analysis, while NELIBS spectra, analyzed with the same technique, provide acceptable results, but with 10% of the spectra not classified. These findings are particularly interesting given the application of the LIBS technique in investigating natural samples having porous and/or rough surfaces. Display omitted •A number of sedimentary and igneous rock samples were analyzed via LIBS and NELIBS.•The NELBS spectral lines intensity showed pronounced improvement compared to LIBS.•Surface hardness of the samples was compared from the spectroscopic data.•PCA and the graph theory techniques were used to characterize the rock samples.•LIBS with graph theory revealed better classification results than NELIBS for rocks.