To solve the load balancing scheduling problem of a reentrant hybrid flowshop （RHFS）, a mathematical RHFS model was formulated, and the weighted summation of the processing time load balancing cost and the total parallel machine waiting time was put as an index for comprehensive evaluation of load balancing. Furthermore, a new en- coding method based on job processing procedure was designed, coupled with time-window constraints and the lar- gest remaining time rules, to finish the decoding process, and a dynamic self-adaptive differential evolution （DSADE） algorithm was used to complete the global optimization. The DSADE algorithm presents a new population update mechanism on the basis of a self- treme hamming dynamic distance to increase the diversity of population, and brings in adaptive parameter adjusting strategy along with stop iterations to enhance the ability to jump out of local ex- value. Finally, an example of production scheduling problem for multi-pass color strip procedure in bus man- ufacturing painting workshop was simulated. The results showed that the load balance evaluation index of the DSADE algorithm was decreased by more than 20% in average compared with the algorithms of GA, differential evolution （DE） and solf-adaptive differential evolution （SADE）.