We show that optimization of Traveling Wave Tube (TWT) design with respect to many design parameters can be efficiently performed by gradient based methods. The partial derivatives with respect to design parameters of various TWT figures of merit can be efficiently calculated by using the adjoint approach [1] , [2] . Several practically important figures of merit have been studied, include average gain, gain flatness, and gain-bandwidth product, and design parameters include beam voltage and circuit geometry. The derivatives are calculated by modified 1D large signal simulation code CHRISTINE-Z [3] and used in the steepest descent optimization algorithm that finds parameter values that minimize or maximize the desired figure of merit. Only three runs of the modified code are needed to compute the partial derivatives of the output power and phase at a specified frequency with respect to an arbitrary number of parameters. It results in a potentially large savings in computing time compared with direct, finite difference calculation of the partial derivatives. illustrate the method by optimizing the beam voltage and gap spacing of a W-band folded-waveguide TWT. The examples of TWT design optimization for small signal gain and output power in large signal regime will be presented and discussed.