Abstract: Diversity loss and low optimizing efficiency are the two problems to be solved for the differential evolution (DE) algorithm in dynamic environment. A multi-population DE algorithm usually applied to the space optimization of continuous solution is proposed for the solution of dynamic scheduling problem with sequential coding, where the representative method using random keys for coding is introduced to transform the continuous position vectors into sequential coding. A self-organizing multi-population strategy is then set out to divide the population into parent population and child population, which is separated automatically from the parent one in accordance to their dynamic spatial characteristics. With the parent population in uninterrupted search of the regions where the peaks take place, the child population is assigned to exploit the useful regions further with adaptive adjustment done for its size, thus expediting the optimizing speed of the algorithm with computing resource saved. The algorithm proposed has been applied to the dynamic scheduling with variable delivery date, and a satisfactory result is gained.