High‐frequency ultrasound (HFU) offers great potential for in vivo imaging of developing mouse embryos. Nevertheless, mouse‐embryo imaging remains challenging because of the random orientation of the embryos and of the limited acoustic penetration depth of HFU. In this study, a 34‐MHz, five‐element annular array was excited using conventional mono‐cycle and chirp‐coded excitations. Synthetic focusing and pulse compression (for the chirp excitation) were used to form images with increased depth of field and penetration depth (for the chirp excitation). Three‐dimensional (3‐D) datasets were acquired from 11‐, 12‐, and 13‐day‐old mouse‐embryo heads using both excitation schemes and a respiratory‐gating algorithm that limited motion artifacts. 3‐D reconstructions of the brain ventricles were formed and compared to MRI‐based reconstructions. The chirp‐based reconstructions were more morphologically accurate than the mono‐cycle‐based reconstructions. Furthermore, only chirp datasets provided sufficient penetration depth to correctly segment the brain ventricles lying deep inside the mother. Differences between the calculated brain‐ventricle volumes obtained using chirp and mono‐cycle excitations were as high as 80%. The results of this study suggest that high‐frequency chirp‐annular‐array imaging could become a valuable tool for translational studies and non‐invasive efficient phenotyping of the early mouse central nervous system. [Work supported by NIH Grants EB006509 and EB008606.]