Haptic interfaces are a class of robotic manipulators that display force feedback to enhance the realism of virtual environment displays. However, these manipulators often fail to effectively replicate the real world environment due to dynamic limitations of the manipulator itself. The ratio of the simulated to transmitted environment impedance is defined as the transparency transfer function (TTF), and can be used to quantify the effectiveness of a haptic device in displaying the simulated environment. The TTF is ideally equal to one for the bandwidth of human proprioception. In this work, a method is presented that increases TTF bandwidth via cancellation of dynamics with an adaptive model. This adaptive model is based on the kinematics and dynamics of a PHANToM haptic interface with assumed joint stiffness and damping added. The Lagrangian of the PHANToM is reformulated into a regressor matrix containing the state variables multiplied by a parameter vector. A least-squares approach is used to estimate the parameter vector by assuming that errors in force output are due to the manipulator dynamics. The parameter estimate is then used in the original model to provide a feed-forward cancellation of the manipulator dynamics. Software simulation using data from passive user interactions shows that the model cancellation technique improves bandwidth up to 35 Hz versus 7 Hz without compensation. Finally, this method has a distinct advantage when compared with other compensation methodsfor haptic interactions because it does not rely on linear assumptions near a particular operating point and will adapt to capture unmodeled features. Copyright © 2005 by ASME.