A rail-to-rail Miller compensation method without feed forward path includes forming a first compensation branch including a first amplifier, wherein an input of the first amplifier is electrically connected with an output of a second stage gain amplifier, the second stage gain amplifier is electrically connected the first stage gain amplifier in series forming an operational amplifier; and forming a second compensation branch including a second amplifier, wherein a dual relation is provided between an input stage of the first amplifier and that of the second amplifier, namely, if the input stage of the first amplifier is N-type, the input stage of the second amplifier is P-type, and vice versa. The present invention is capable of achieving the rail-to-rail output range without affecting the system stability. The N-type and P-type inputs are simultaneously applied to the input of the amplifier of the compensation branches.