This paper presents an adaptive teleoperation which is robust to time-delay and environmental uncertainties while assuring the transparent performance. A novel theoretical framework and algorithms for the teleoperation system have been built up with neural network-based multiple model control and time forward state observer. Conditions for stability and transparency performance are also identified. Simulations show that the system is stable and in good performance. © 2010 Vu Trieu Minh and Fakhruldin Mohd Hashim.