The quality of maintenance schedule has a great impact on the security and economy of distribution system. A new model for distribution maintenance scheduling is proposed， which combines load transfer and maintenance schedule optimization problem. In this model， the maintenance schedule optimization problem is the main problem and the other is the sub-problem. By the solving reiteration of main and sub-problem， it finally gets an optimal maintenance schedule with the least electricity sell loss and an optimal load transfer plan with the least load loss， switch operations and transmission loss. To solve this complicate problem，this paper uses a hybrid strategy that mixed with Immune Algorithm（IA） and Tabu Search（TS） to solve main problem and improves basic candidate restore tree cutting algorithm. By applying connect switch choosing rules， it can decrease the candidate restore tree cutting number and accelerate calculating speed. By adding a network loss comparative tache in the solve process， it can obtain a most optimal solution with least network loss. The proposed model and method is applied to a test system and numerical results verify the correctness and validity.