We propose a sensor selection method for time difference of arrival and angle of arrival localization scenario. When the number of available sensors is given, we select the combination which has the best localization accuracy. The localization accuracy can be described by the Cramer-Rao Lower Bound but we cannot attain it in practical applications. Hence, we introduce the closed-form solution of time difference of arrival and angle of arrival and give the localization error covariance matrix. Theoretical analysis and simulation show that the performance of the proposed closed-form solution can attain the Cramer-Rao Lower Bound. We use the localization error covariance matrix of the proposed method to describe localization accuracy. An optimization problem is formulated to determine which sensor combination should be selected and we convert this non-convex problem into convex problem by relaxing the constraint condition. The simulation results validate that the proposed sensor selection for time difference of arrival and angle of arrival localization is effective.