Objective To develop a practical image acquisition strategy using intermittent breath?hold cone beam computed tomography (CBCT). Methods A breathing phantom was used to simulate the movement of tumor near the diaphragm during free breathing and breath hold and scanned by conventional breath?hold CBCT and type Ⅰ/Ⅱ intermittent breath?hold CBCT. In the conventional breath?hold CBCT, scan paused and free breathing occurred at the break of breath hold and free breathing was not included in the scan. In the intermittent breath?hold CBCT, one scan covered several breath holds separated by free breathing in a ratio of 3 vs1. Image quality and three?dimensional registration accuracy were quantitatively compared between conventional breath?hold CBCT and type Ⅰ/Ⅱ intermittent breath?hold CBCT. Comparison of image quality parameters between conventional breath?hold CBCT and intermittent breath?hold CBCT was made by paired t test. Results Motion artifacts arose in type I and Ⅱ intermittent breath?hold CBCT scans. There were no significant differences in the reconstructed pixel value or uniformity between intermittent breath?hold CBCT and conventional breath?hold CBCT ( P>0. 05, and P= 0. 02, 0. 53 ) . Compared with conventional breath?hold CBCT images, the signal?to?noise ratios of type I andⅡintermittent breath?hold CBCT images were reduced by 30% and 60%, respectively ( P<0. 05 ) . The registration error was up to 0 . 4 cm in the anterior?posterior direction and less than 0 . 1 cm in other directions . Conclusions The phantom study shows that intermittent breath?hold CBCT does not significantly reduce image quality or registration accuracy compared with conventional breath?hold CBCT. The feasibility of intermittent breath?hold CBCT in clinical application needs to be further validated among a large number of patients.