Industrial process tomography (PT) possesses unique advantages in multiphase distribution imaging. Oil–gas–water three-phase distribution has coexisting high-impedance contrast and low-impedance contrast medium, possessing a challenge to traditional PT sensing technique. An original concept of multifrequency ultrasonic tomography (MFUT) is proposed, which reconstructs oil–gas–water three-phase distribution with the single modality of UT. Operated in the transmissive ultrasonic tomography (UT) framework, the proposed MFUT quantifies attenuation spectrum response under frequency-modulated chirp signal excitation and extracts fluid absorption coefficient through partial differential modeling of frequency attenuation response. Artifacts introduced by lateral scattering is further alleviated through beam angle modification during assembling of the sensitivity matrix. Thus, the MFUT method is theoretically less sensitive to nonlinear error introduced by the reflection/refraction/scattering effect. In numerical and experimental results, quantitative attenuation characterization offers MFUT the capability of reconstructing gas–water/oil–water/oil–gas–water multiphase distribution with higher accuracy, boundary preservation, and image purity. Synthetic modeling of spectral attenuation also outperforms the single-frequency image fusion and frequency difference reconstruction, from which the MFUT is proven feasible to investigate a multiphase system.