Abstract: Aiming at the scheduling problem of fuzzy hybrid flowshop with parallel machines, this paper proposes an improved distribution estimation algorithm (IEDA). In the proposed algorithm, the method of ranking fuzzy numbers is used to establish the scheduling model, and the minimization of the weighted average and uncertainty of the fuzzy makespan is taken as the objective of scheduling. An initial population is generated by means of the NEH(Nawaz Enscore Ham) heuristic and the strategy of destruction and construction. The variable neighborhood searching is incorporated to enhance the local exploitation, and the strategy of destruction and construction is applied to improve the diversity of population. Moreover, when the best solution has no improvement for successive generations, the variable neighborhood searching will be adopted so as to escape from local optimum. In addition, an orthogonal experiment design is utilized to adjust the parameters of IEDA. The simulation results indicate the superiority of the proposed IEDA.