This paper presents the balancing control of a unicycle robot using air power. Since the robot has one wheel to move forward and backward, the balancing control is quite challenging. To control the balancing angle, the accurate angle estimation by a tilt and a gyro sensor is required a priori. A complementary filter is implemented to eliminate the defects of two sensors and to fuse together to estimate an accurate balancing angle. The optimal design of air ducts is found empirically. Experimental studies of the balancing control of a unicycle robot confirm that the robot is well regulated without falling down.