Tabu (or taboo) search (TS) has been successfully applied to combinatorial optimization but it has not been used for global optimization of many continuous functions including phase equilibrium calculations via Gibbs free energy minimization. In this study, a version of TS, namely, enhanced continuous TS (ECTS), is first tried for benchmark test functions having multiple minima and then evaluated for phase equilibrium calculations. Examples for the latter involve several components, typical conditions and common thermodynamic models. Performance of ECTS is compared with a genetic algorithm (GA). The results show that both TS and GA have high reliability in locating the global minimum, and that TS converges faster than GA thus reducing the computational time and number of function evaluations. (C) 2003 Elsevier Ltd. All rights reserved.