A cap opening system which can handle various containers having caps and container bodies of different sizes is disclosed. In this system, when a cap of a container is to be opened, a container body of the container is gripped by a container body holding apparatus to raise the container body upward. A top surface of the cap is detected as a reference surface when the cap interrupts an optical beam. Based on the height of the reference surface, the cap is positioned with respect to the cap handling apparatus. Alternatively, a cap receiving member is provided underneath the cap handling apparatus to perform positioning of the cap with respect to the cap handling apparatus.