Space restrictions prevent surgeons to directly interact with the patient during magnetic resonance imaging (MRI)-guided procedures. One practical solution would be to develop a robotic system that can act as an interface between surgeon and patient during those interventions. The proposed system consists of a commercial PHANTOM device (product of The Sensable Technologies) as the master robot and an MRI-compatible patient-mounted parallel platform (that we name ROBOCATHETER) designed to serve as the slave mechanism inside the scanner bore. As the main contribution of this paper, a linear parameter varying (LPV) gain-scheduling controller is designed and implemented to obtain the desired performance of the slave robot in tracking set points and reference trajectories. To do so, a reduced-order dynamic model of the robot based on the Lagrange method is derived to capture the nonlinear dynamics of the platform. The model is then used for the design of an output-feedback LPV controller to command the robot to position the catheter in any desired states. During the course of control, appropriate selection of scheduling parameters not only helps to compensate for the nonlinearities of the system dynamics but also leads to a set of decoupled models for the system, where each degree-of-freedom (DOF) could be treated separately. The performance of the controller is compared with a variable-gain proportional-derivative-integral (PID) controller. Experimental results show that the proposed control scheme has significant advantages in terms of set point tracking and actuator saturation over the baseline PID controller.