A procedure for the estimation of the number of peaks in a chromatographic
peak cluster is proposed. This procedure is based on the dependence of the
sum of squared errors (differences) between real and calculated chromatogra
ms on the considered number of overlapped peaks. Starting with the number o
f apparent peaks, chromatograms are deconvoluted successively with increasi
ng numbers of peaks possibly present. The sum of squared errors decreases w
ith increasing numbers of peaks up to a point beyond which the function bec
omes constant. This point corresponds to the best estimation of the real nu
mber of peaks present in a cluster. The procedure was successfully used for
the peak number estimation in model and real chromatograms.