A method for displaying caption windows in a TV system is provided. The method includes steps of receiving a first window-defining command at a first time point, receiving a second window-defining command at a second time point, and determining an output state of the window-defining commands according to a comparison result between a first global priority value corresponding to the first window-defining command and a second global priority value corresponding to the second window-defining command. The first global priority value includes a first priority value of the first window-defining command and a first age parameter. The second global priority value includes a second priority value of the second window-defining command and a second age parameter. The first and second age parameters are associated with a sequence of the first and second time points.