The study investigated the algae growth and lipid change by using confocal Raman microscope, and the Raman spectra were obtained from the Chlorella pyrenoidosa growing under three nitrogen conditions (nitrogen deficiency, normal and excess). Bubble diagrams of the ratio of lipid characteristic peak were presented as an intuitive expression of lipid accumulation, which was corresponding to the NR fluorescence image in some way. The preprocessed Raman signals were analyzed by using principal components analysis. Linear discriminant analysis (LDA) was used to establish a classification model by using appropriate principal component variables. The prediction accuracy obtained from LDA prediction model established by the three nitrogen conditions were 80%, 93.3% and 86.7%, respectively. The LDA classification model was established by lipid-related Raman shift (RS) at the three nitrogen conditions, and its prediction accuracy reached to 86.7%. The results showed that the identification of different nitrogen stress on microalgae grown using Raman technology was feasible, and with time passing by, the difference in the accumulation of lipid became greater.