In this paper, we propose a Pareto-based tabu search algorithm for multi-objective FJSP with Earliness/Tardiness (E/T) penalty. In the hybrid algorithm, several neighboring structure based approaches were proposed to improve the convergence capability of the algorithm while keep population diversity of the last Pareto archive set. In addition, an external Pareto archive was developed to record the non-dominated solutions found so far. In the hybrid algorithm, dynamic parameters were introduced to adapt to the searching process. Experimental on several well-known benchmark instances show that the proposed algorithm is superior to several existing approaches in both solution quality and convergence ability.