Summary: The machine-part cell formation with respect to multiple objectives has been an attractive search topic since 1990 and many methodologies have been applied to consider simultaneously more than one objective. However, the majority of these works unify the various objectives into a single objective. The final result of such an approach is a compromise solution, whose non-dominance is not guaranteed. A Pareto-optimality-based multi-objective tabu search (MOTS) algorithm is presented for the machine-part grouping problems with multiple objectives: it minimizes the total cost, which includes intra- and inter-cell transportation cost and machine investment cost, minimizing the intra-cell loading unbalance and minimizing the inter-cell loading unbalance. A new approach is developed to maintain the archive storing non-dominated solutions produced by the tabu search. The comparisons and analysis show that the proposed algorithm has considerable promise in multi-objective cell design.