To solve the min-max vehicle routing problem（ MMVRP）

aiming at shortening length of the longest sub-line

an elite opposition-based learning artificial fish swarm algorithm（ AFSA） is proposed. According to the customer’s nearest neighbor matrix

the initial artificial fish swarm is constructed to improve the quality of the fish swarm. Considering the features of this problem and the artificial fish swarm algorithm

the intelligent behaviors of artificial fishes are redefined. Inspired by opposition-based learning

two mechanisms are built for constructing discrete opposition-based solutions including full elite opposition-baesd learning and partial elite opposition-baesd learning. Simulation results show that the elite opposition-based fish swarm algorithm can get a high search quality and stability in solving this problem.