Several traditional stopping criteria in Genetic Algorithms (GAs) are applied to the optimization process of a typical laminated composite plate. The results show that neither of the criteria of the type of statistical parameters, nor those of the kinds of theoretical models performs satisfactorily in determining the interruption point for the GA process. Here, considering the configuration of the history curve of the maximum objective or fitness value with generation number, a Logarithmic Performance Criterion (LPC) is introduced as a stopping criterion for optimization of composite plates. The results highlight the advantages of LPC in performing sufficiently smooth (no noise), requiring reasonable number of generations, less parameter dependency, lower need for conservative assumption, on-line controllability, wide scope applicability, and reasonably easy application in engineering decision-making problems.