The adaptive region control method of the pitch and roll attitudes for operating autonomous underwater vehicle (AUV) is addressed. In the actual operation process, there are some influencing factors such as the operation disturbance of manipulator and environmental uncertainty, which will affect the attitude control of the underwater vehicle, and then affect the movement and operation accuracy. To solve the attitude stability control problem, an adaptive region control method of AUV based on radial basis function (RBF) neural network is proposed. An RBF neural network is applied to estimating the uncertain factors of the dynamics model online, and the sliding mode control term is incorporated to compensate the estimation error online. Then, an online adjustment law of the network weights, the center of the radial basis function and the variance is designed to solve the value of the control parameters of RBF neural network, and self-adaptive learning of the control parameters is achieved to adapt to the attitude changes of the vehicle. Moreover, a PI (proportional-integral) section is added to the region controller to realize the fast and stable convergence of the attitude, and thus the adjustment time and steady state error are reduced. Based on the Lyapunov stability analysis, it is theoretically proved that the region control error is uniformly asymptotically stable. Finally, the pitch and roll attitude control experiments of the operating AUV prototype are conducted to verify the effectiveness of the proposed method.