The region of interest (ROI) encryption in the video can reduce the complexity of calculation and improve encryption speed by encrypting only the area containing critical visual information. Above all, ROI encryption can expand the utilization domain (e.g., video surveillance), unlike general encryption methods that de-identify the entire frame. However, the traditional ROI encryption process in high-efficiency video coding (HEVC)/H.265 is more complex than in advanced video coding/H.264, and the encrypted area tends to be wider than the ROI. Thus, cryptographic algorithms are applied outside the ROI, which wastes computing resources and restricts the visual information that needs to be provided. Therefore, this paper proposes a coding unit (CU)-based ROI encryption for HEVC/H.265 video. The proposed method selectively encrypts HEVC/H.265 parameters, such as the intra prediction mode, motion vector (MV), MV sign, transform coefficient (TC), and TC sign, which have significant visual influence, and adopts the tile concept for parallel processing frames. The CU-based ROI encryption reduces complexity by identifying the encryption area based on the CU coordinates, applying encryption only to the CUs associated with the ROI. This approach can preserve the area around the ROI by restricting the reference area. Moreover, it provides up to about 30% faster encryption speed than the traditional method while maintaining performance (i.e., peak signal-to-noise ratio and structural similarity index measure).