Abstract: For building respiratory signal acquisition platform, the respiratory signal obtained by flow-sensor was denoised using wavelet analysis. The relation between selection of wavelet function, decomposition order and threshold and the quality of wavelet filtering was analyzed through simulation, and the optimal wavelet function and decomposition order was obtained. Lastly, respiratory signal was acquired and filtered by the executable-file running independent of Matlab successfully, which was realized by the mixed program between VC++ and Matlab utilizing COM technology of Matlab, and the simulation results was better.