The important features of sound spectra of nasal consonants have been examined quantitatively with the assistance of a computer. The computer program (1) allows the experimenter to select the frequencies and bandwidths of poles and zeros of an arbitrary transfer function, (2) calculates the corresponding sound spectrum, (3) provides visual comparison between the calculated spectrum and any given stored speech spectrum, and (4) calculates measures of the goodness of fit of the two spectra. The speech samples were drawn from the materials described in a paper on vowel formants presented at this meeting. In order to match spectra of nasal consonants in the frequency range below 3000 cps, one or two pole‐zero pairs attributable to the closed oral cavity must be used, together with three or four poles. The bandwidths of the poles in nasal consonants are always greater than the bandwidths found in nonnasal vowels. The locations of poles and zeros which give the best match for [m] may vary significantly with vowel environment and may change during the period of mouth closure. Theoretical considerations can explain and to some extent predict these pole‐zero characteristics. [This work was supported in part by the U. S. Army (Signal Corps), the U. S. Air Force (Office of Scientific Research, Air Research and Development Command), and the U. S. Navy (Office of Naval Research); and in part by the National Science Foundation.]