Reliable spectral analysis is only achieved if the spectrum is thoroughly investigated in regard to all hidden and overlapped peaks. This paper describes the steps undertaken to find and separate such peaks in the range of 3000 to 4000 cm(-1) in the case of three different infrared absorption spectra of the glass surface of hydrolyzed silica optical fibers. Peak finding was done by the analysis of the second and fourth derivatives of the digital data, coupled with the available knowledge of infrared spectroscopy of silica-water interaction in the investigated range. Peak separation was accomplished by curve fitting with four different models. The model with the best fit was described by a sum of pure Gaussian peaks. Shoulder limit and detection limit maps were used to validate the revealed spectral features.