In order to discriminate and identify phytoplankton of different divisions and genuses, coiflet2 (coif2) wavelet function was utilized to extract the characteristics of the three-dimensional (3D) fluorescence spectra of 12 phytoplankton species belonging to 9 genuses of 4 divisions. The third scale vectors selected as the discriminating characteristic spectra, obviously express the distinguish characteristics of different genuses and divisions. The results of Bayes discriminant analysis showed that these characteristic spectra had average discriminating rates of 99.0% and 97.4% at the division and the genus level, respectively. Reference spectra were obtained from these characteristic spectra by cluster analysis. A fluormetric method was established by multiple linear regression resolved by the nonnegative least squares. These reference spectra identified the single species added with 10% and 20% ratios of random noise with the rates of more than 98.0% and 85.0%, respectively, at the division and the genus level. All the dominant species of the phytoplankton mixtures could be identified 100% at both the division and the genus level.