A method is proposed for analyzing the relative energy distributions of boundaries in ultrafine-grained materials measured by grain boundary grooving using scanning tunneling microscopy. The energy distribution of boundaries in a grain boundary ensemble is considered as a superposition of individual distributions (populations), which can be identified by cluster analysis based on statistical criteria and each of which has its own value of average energy, dispersion, and share in the overall distribution. The energy distributions of grain boundaries are analyzed for 12Cr15Mn9NiCu steel with a coarse-grained structure in the as-received state and with an ultrafine-grained structure produced by hot helical rolling and subsequent cold rolling. It is shown that the number of boundary populations and their main characteristics revealed by cluster analysis depend on the steel structure. The results of cluster analysis of experimental distributions are compared with the EBSD measurement data on grain boundary misorientation distributions. Discrepancy between the cluster analysis results for the energy and misorientation distributions of boundaries is discussed taking into account the difference in the nature of the data obtained.