This paper discusses the design of an adaptive control system for robotic-assisted surgery with haptic feedback. Through a haptic device, the surgeon tele-operates the medical instrument, fixed in a remote robot, in free space or in contact. In free space, the surgeon feels the motion of the robot. When contact with tissues occurs, an on-line environment stiffness estimation is performed. Without considering a contact position, the environment estimation parameter strategy uses an Extended Kalman Filter with robot modeling errors compensation. Based on the transparency's frequency analysis, the estimated parameter is used to adapt the teleoperation scheme to ensure stability and to improve telepresence.