Various abdominal compression methods using some mechanical devices have been reported to suppress tumor and organ motions during irradiation. In particle therapy, however, it is desirable not to place such a device near the irradiation field to prevent deterioration of dose distribution due to scattering. Therefore, we developed a device-free compressed shell fixation method and have treated liver tumors using this system. This study examined the effect of suppressing respiratory motion of the liver by this method. This was a part of clinical studies for primary hepatocellular carcinoma and metastatic liver tumor using proton beam therapy based on protocols elaborated at Nagoya Proton Therapy Center and approved by the institutional review board. 146 patients (102 males and 44 females) with fiducial markers placed in the vicinity of the target tumor prior to proton beam therapy were included in this study. The patients were fixed to the couch using a thermoplastic shell made of polycaprolactone resin (ESFORM; Engineering System Co., Ltd., Nagano, Japan). When this shell was shaped to fit the patient’s body, a beanbag weighing 2, 3, or 5 kg was placed on the epigastrium of the patient for the purpose of suppressing the respiratory movement of the liver. After the shell cooled down and the beanbag was removed, the device-free compressed shell was completed. After making the shell, the patient was instructed to breathe about 14 times per minute, and finally 4D-CT for treatment planning was taken. The magnitude of respiratory movement of the liver was calculated by analyzing the coordinates in the cranio-caudal direction of the fiducial marker from the X-ray fluoroscopic images at the time of shell molding and 4DCT for treatment planning. The magnitude of suppression indicating the overall effect of the shell fixation and breathing method instruction was calculated by subtracting the magnitude of respiratory movement at the 4D-CT from the value before making the shell for each patient. The median respiratory movement before making the shell, after fixing with the shell, and after instructing the breathing method were 14.2 mm (interquartile range (IQR), 10.7-19.8), 11.5 mm (IQR, 8.6-17.5), and 10.4 mm (IQR, 7.3-15.8), respectively. These values were significantly different in Kruskal-Wallis test. The median magnitude of suppression was 3.9 mm (IQR, 1.3-7.8). There was no significant difference in the magnitude of suppression according to the weight of the beanbag, sex, and BMI. The patient group wearing a shell made with the 5-kg beanbag had a significant variability in the value in Mann-Whitney U-test. The device-free compressed shell fixation method was efficient in suppressing respiratory movement of the liver. This method requires no device for abdominal compression, so compatibility may be good, especially for particle beam therapy.