A method for analyzing a multichannel chromatogram is realized for accurately resolving overlapping peaks on a multichannel chromatogram to permit analysis of the composition of a sample. First, components of overlapping peaks are specified. A data matrix Dij is determined, and compressed in the wavelength direction to obtain a data matrix vector Di. By compressing three-dimensional data in the wavelength direction, noise components cancel with each other to obtain noise free two-dimensional data Di.beta.. Next, the two-dimensional data is deconvoluted, final standard deviation s0 is registered, and the deconvoluted two-dimensional data is reconvoluted to a retention intensity matrix. A quantitative spectral intensity matrix is computed, components are identified, and the identified components are quantified. The quantified data is displayed on a screen. Since noise components cancel with each other, it is possible to realize an analyzing method which is capable of accurately resolving and analyzing the components. Also, overlapping peaks are specified, and a data matrix D is determined. Subsequently, an eigenvalue problem is solved, and the number n of components is estimated. The deconvolution and computation of the eigenvalue problem are iteratively executed until the peaks are isolated. Then, elution profiles are calculated, the components are identified, and the identified components are quantified.