This paper describes task-oriented manipulability of teleoperated robotic arms mounted on remotely operated vehicle (ROV) and its application to task-oriented joint configurations. Main purpose of the study is to reduce teleoperator's burden in performing underwater tasks by enhancing the functionality of manipulator. Even though a manipulator has 6 degrees-of-freedom (DOF), which is proper DOF to work in Cartesian workspace, the manipulator might have redundancy according to task types and order of task-priority. This paper focuses on the problem to utilize the redundancy by introducing a scalar function as an object of optimization. The scalar function is composed of task-oriented manipulability measure (TOMM) and joint limit measure (JLM). Using sequential quadratic programming (SQP) with the object function, we obtained optimal postures of manipulator for a given position constraint of end-effecter. Adopting the scalar function as a performance index, we solve a redundancy resolution problem based on pseudo inverse of task-oriented Jacobian matrix