Abstract: The objects adopted in the present study were adult male birds from 10 oscine species obtained either from wild field or from breeding in our laboratory.When birds were acquainted with their surrounding livings,songs were recorded,and analyzed by use of VS-99 sound analyzing software.Song complexity was valued through the following seven index:total phrase (TP),syllables in all the phrases (SAP),the syllable types in all the phrases (STAP),syllables in the longest phrase (SLP),the syllable types in the longest phrase (STLP),the average number of syllable per phrase (ANSPP) and the average number of syllable types per phrase (ANSTPP).After the recordings finished,we measured the volumes of three song control nuclei:HVC (high vocal control center),RA (robust nucleaus of the archistriatum) and Area X.Then,the seven index assessing the song complexity and the volumes of HVC,RA and Area X were clustered by cluster analysis software (SPSS 10.0,significant level=0.05).The results revealed that song complexity largely varied not only among different families but also among species in the same family.Three studied nuclei were the largest in the Mongolian lark (Melanocorypha mongolica),then in the Eurasian siskin (Serinus canaria) and yellow-throated bunting (Emberiza elegans).The results of cluster analysis based on the song complexity and the size of song nucleus differed from each other,and both of them were different from those of traditional classification.However,according to the cluster analysis of RA and Area X volumes,the Mongolian lark could be distinguished from the others,which was consistent with the classification based on the traditional or DNA data classification.