Purpose: In radiotherapy that takes into account respiration using a RPM (Real time Position Management, Varian, USA) system, which can treat in consideration of the movement of tumor, infrared reflective markers supplied by manufacturers cannot obtain respiratory signal if the couch rotates at a certain angle or larger. In order to solve this problem, the author developed the 3D infrared reflective marker named 'Kw-marker' that can obtain respiratory signal at any angle, and evaluate its usefulness. Materials and Methods: In order to measure the stability of respiratory signal, we put the infrared reflective marker on the 3D moving phantom that can reproduce respiratory movement and acquired respiratory signal for 3 minutes under each of 3 conditions (A: $couch\;0^{\circ}$, a manufacturer's infrared reflective marker B: $couch\;0^{\circ}$, Kw-marker C: $couch\;90^{\circ}$, Kw-marker). By analyzing the respiratory signal using a breath analysis program (Labview Ver. 7.0), we obtained the peak value, valley value, standard deviation, variation value, and amplitude value. In order to examine the rotation error and moving range of the target, we placed a B.B phantom on the 3D moving phantom, and obtained images at a couch angle of $0^{\circ}$ and $90^{\circ}$ using OBI, and then acquired the X, Y and Z values (mm) of the ball bearing at the center of the B.B phantom. Results: According to the results of analyzing the respiratory signal, the standard deviation at the peak value was A: 0.002, B: 0.002 and C: 0.003, and the stability of respiration for amplitude was A: 0.15%, B: 0.14% and C:0.13%, showing that we could get respiratory signal stably by using the Kw-marker. When the couch rotated $couch\;90^{\circ}$, the mean rotation error of the ball bearing, namely, the target was X: -1.25 mm, Y: -0.45 mm and Z: +0.1 mm, which were within 1.3 mm on the average in all directions, and the difference in the moving range of the target was within 0.3 mm. Conclusion: When we obtained respiratory signal using the Kw-marker in non-coplanar treatment where the couch rotated, we could acquire respiratory signal stably and the Kw-marker was effective enough to substitute for the manufacturer's infrared reflective marker. When the rotation error and moving range of the target were measured, there was little difference, indicating that the displacement of the reflector movement in couch rotation is the cause of change in the scale and amplitude of respiratory signal. If the converted value of amplitude height according to couch angle is studied further and applied, it may be possible to perform non-coplanar phase-based gating treatment.