The high permeability and uncertainty of distributed photovoltaics have a great impact on the node voltage and power flow distribution and pose a new challenge to the safe operation of distribution network. How to effectively and accurately optimize and control distribution network containing distributed photovoltaics is an urgent problem to be solved. A model predictive control optimization method for distribution network containing distributed photovoltaics is proposed. In order to reduce the voltage deviation of nodes and power loss, the prediction information is continuously updated and the optimization model is constructed based on the model predictive control, and the rolling optimization control of distributed photovoltaics and energy storage devices is carried out. A modified IEEE 33-node system is used as an example to verify that the voltage deviation of nodes and power loss of distribution network are effectively reduced. The proposed method has the better optimization effect than the day-ahead optimization method does for fitting the actual resuts. Based on the model predictive control, the proposed method is helpful to the safe operation of distribution network under the distributed photovoltaics access, thus reducing the deviation between the control dispatch and the actual situation.