Stability or performance of robot control system can be reduced because of nonlinearity, strongly coupled dynamic and modeling uncertainty such as unknown load mass, joints inertia and friction parameters. In this thesis, the problem of robot manipulator control is investigated in the conditions that no specified and accurate dynamic model of the system is available. In other words, the model has a structured uncertainty for example in the inertia matrix or gravity vector. The aim is to propose a control law for robust trajectory tracking of robot manipulator. Nonlinear and robust structure of sliding control caused this approach to be an ideal candidate for this application. Although appropriate features of sliding approach, such as applicability for linear and nonlinear process, insensitivity to the parameters variation and external disturbance, good transient response and simple implementation are available, this approach has some defects. For example, non-ideal sampling and switching may cause high frequency chattering in the system, which in turn can excite un-modeled high frequency dynamics and degrade the system stability. Furthermore, the chattering increases controller burden and may easily damage controller parts. Tracking error reduction in addition to chattering alleviation is the main concern of this thesis. This thesis presents a new control approach based on sliding mode theory for robot manipulator. The presented control law includes a multivariable exponential function to eliminate the control signal chattering. Through some theorems, convergence of the states to the sliding surface and uniform global asymptotic stability of the proposed control system are guaranteed based on Lyapunov stability theorem for non-autonomous systems.within the proof a new Lyapunov function has been presented. The proposed approach decreases the tracking error while improves the system speed response and presents a satisfactory control performance as well. Some of the inter-ideograph; TEXT-ALIGN: justify; MARGIN: 0cm 0cm 0pt" Keywords Chattering, Lyapunov stability, Multivariable control, PUMA 560, Robot manipulator, Sliding control, Tracking