The objective of our research was to predict cell fates of a multicellular system, accompanied by cellular differentiation. To fulfill this objective, we sought to distinguish the differentiated and undifferentiated cells of filamentous cyanobacteria (Anabaena sp. PCC 7120) using Raman imaging. This technique indicated Raman bands of the cellular system, in which several bands were assigned to vibrations of β-carotene and scytonemin. We applied principal component analysis (PCA) to the Raman spectra to determine the PC1 and PC2 loading plots and their scores. The data points obtained for heterocyst tended to converge along the bottom of the scatterplot whereas those for vegetative cells were more widely distributed in the PC plane. This indicates that the chemical compositions of a heterocyst were relatively stable. As vegetative cells are capable of proliferation or differentiation, they may transit and exist in several states including the pseudo-differentiated state. The results suggest that the chemical compositions of a vegetative cell fluctuated according to its cellular condition. In conclusion, the results of Raman imaging indicate that the diverse states of vegetative cells are localized in a specific state through differentiation.