Wolf is a threaten species in many countries around the world and wolf howling is one of the main monitoring tools. This study aimed at improving wolf howling technique using spectrographic analysis to count wolves and to assign howls to the correct pack or individual. Chorus and solo howls from 6 different free ranging packs were elicited by playback from 2003 to 2007 on the Tuscany Apennines (Italy); in all, 127 howls have been analyzed. Minimum number of wolves has been estimated considering contemporaneous howls in the chorus. We replied the analysis in different points of the sound recordings to limit bias due to possible biphonation and other non linear phenomena. Discrimination was performed by discriminant function analysis. We censed 27 animals (mean 4.5 individuals/pack). Cross-validated discriminant analysis assigned 75.7% of the howls to the correct individual and 79.7% of the howls to the correct pack. As regards DFA analysis, final and maximum fundamental frequencies are the most important howl features to discriminate among individuals and among packs. Individual discrimination could be a useful tool to validate the number of animals censed by spectrographic visualization. Future investigations will be aimed to identification, thus associating howls with individuals over periods of time, a promising tool for wolf conservation.