Although slow-rotation CT scanning (slow-scan CT: SSCT) has been used for radiation therapy planning, based on the rationale that the average duration of the human respiratory cycle is 4 s, a number of physical and quantitative questions require answering before it can be adopted for clinical use. This study was performed to evaluate SSCT physically in comparison with other scan methods, including respiratory-gated CT (RGCT), and to develop procedures to improve treatment accuracy. Evaluation items were geometrical accuracy, volume accuracy, water equivalent length and dose distribution using the 256-detector row CT with three scan methods. Fast-scan CT (FSCT) was defined as obtaining all respiratory phases in cine scan mode at 1.0 s per rotation. FSCT-ave was the averaged FSCT images in all respiratory phases, obtained by reconstructing short time intervals. SSCT has been defined as scanning with slow gantry rotation to capture the whole respiratory cycle in one rotation. RGCT was scanned at the most stable point in the respiratory cycle, which provides the same image as that by FSCT at the most stable point. Results showed that all evaluation items were dependent on motion characteristics. The findings of this study indicate that 3D planning based solely on SSCT under free breathing may result in underdosing of the target volume and increase toxicity to surrounding normal tissues. Of the three methods, RGCT showed the best ability to significantly increase the accuracy of dose distribution, and provided more information to minimize the margins. FSCT-ave is a satisfactory radiotherapy planning alternative if RGCT is not available.