The applcation of the fast Fourier transform in the digital analysis of various biological signals in the frequency domain is well-known. In this paper a method is described for spectrum analysis of respiratory sound using the technique of fast Fourier transform. Results of the application of the method to respiratory sounds of six normal subjects and six tuberculosis patients are presented. Such studies may well prove to be a promising noninvasive diagnostic tool for lung diseases and increase the clinical importance of lung auscultation.