Objective To compare the effects of different respiratory signal acquisition methods on the delineation of moving tumor targets. Methods A cube phantom containing a sphere was placed on a motion platform to simulate respiratory movement by setting motion period, frequency, and direction. Respiratory signal was acquired by real-time position management (RPM) method and GE method independently. Target delineation was conducted using the maximum intensity projection (MIP) sequence. The difference between the reconstructed volume and the theoretical moving volume was compared under the two respiratory signal acquisition methods for cube and sphere targets. Results Under the same respiratory signal acquisition method, the same respiratory amplitude, and different respiratory frequencies, reconstructed volume changes were relatively small. For the sphere target, the deviation between the reconstructed volume and the theoretical moving volume was −1.5% to 5.7% with the RPM method and −1.3% to −13.8% with the GE method (both P < 0.05). For the cube target, the deviation between the reconstructed volume and the theoretical moving volume was 0.2% to 0.9% with the RPM method and −2.6% to 0.9% with the GE method, with no statistical significance. Conclusion For small-volume sphere targets, the target volumes obtained from MIP images by the two respiratory signal acquisition methods are both smaller than the actual moving volume. For large-volume cube targets, there is no significant difference between the reconstructed and theoretical results with any respiratory signal acquisition method. The RPM method produces smaller deviation and better image quality when reconstructing small-volume targets.