In this paper the effects of the nasal cavity and the paranasal cavities on the spectra of nasal sounds are investigated. The oral output and the nasal output were measured separately as shown in Fig. 1, each of which was used for the spectral analysis. The examples of analyzed spectra of nasalized vowels and nasal murmurs are shown in Fig. 2. The anatomical structures of the nasal cavity and the paranasal cavities are discussed with the anatomical charts (Fig. 4). The physical dimensions of paranasal cavities are discussed with the anatomical charts (Fig. 4). The physical dimensions of paranasal cavities were estimated as shown in Fig. 6 using a human skull (Fig. 5). The area functions of the pharynx, the oral cavity, and the nasal cavity were determined as exemplified in Fig. 8 by applying the data published by Chiba and Kajiyama and by Fant. The frequencies of poles and zeros which specify the transfer function were calculated using the models for the productions of nasal sounds shown in Fig. 7. These calculations were carried out by computer simulation of the well-known theory of the transfer characteristics of acoustic tubes. The spectra were synthesized as shown in Fig. 9 from the frequencies of poles and zeros from simulation. These synthesized spectra are compared with the analyzed spectra shown in Fig. 2, and the nasal cavity and each paranasal cavity are discussed in detail. Especially, the zeros which appear in the low frequency region are discussed with the supplementary examinations of the spectral changes, in which the nostrils were closed as shown in Fig. 12. As the result of the discussions, we may conclude that the spectral characteristics of the nasal sounds in natural utterances cannot be fully explained by the simplified model including only the nasal cavity, and that the effects of the paranasal cavities must be also considered.