The dual-actuator structure for hard disk drives (HDDs) allows faster data access with relatively slower spindle speed. However, the mechanical interaction between the actuators tends to degrade the head positioning accuracy and thus decreases the track density performance. In this paper, the dual-actuator structure is formulated as a dual-input dual-output (DIDO) model and the parameters of the model are identified by least square estimation algorithm in frequency domain. The loop shaping control designs are presented to deal with the narrow-band disturbance due to interaction generated by the other actuator doing random-seeking. Experimental results on a prototype demonstrate that the position error signal (PES) value due to interaction disturbance is reduced by 16% with a notch filter and 64% with a peak filter based on the lead-lag feedback controller.