We proposed a passive method to distinguish and to estimate density of the Cyanobacteria (Cyanophyceae) in a mixed population by measuring the spectral absorption of sample waters, based on two step linear regression analysis. Natural freshwater usually contains a few species of algae and dissolved organic carbon (DOC). In the experiment, we picked out four typical algal groups characterized by their own colors, Cyanophycaeae or blue-green alga, Chlorophyceae, or green alga, and Bacillariophyceae and Dinophyceae or brown algae. In the first step, for each of the pure sample waters which contained only one of these elemental substances, dependence of spectral characteristic on its density was determined using simple linear regression analysis. Resultant spectral characteristics which we call gradient vectors were used to estimate spectral absorption of mixed sample waters containing the four elementary algae and DOC by multiple linear regression analysis. This method offers new perspectives for identification and estimation of density of blue-green algae and other unialgal species in a mixed population. (c) 2005 The Optical Society of Japan.