This paper reports our investigations on the design of globally or quasi-globally optimal structures for three-component and four-component mechanically compensated zoom lenses. This is accomplished by implementation of a global optimization technique based on evolutionary programming. The technique searches optimal structures in the configuration space formed by the specific design variables: powers of individual components and the intercomponent separations. Any requirements for system length and Petzval curvature of the zoom lens can be incorporated in the search for optimal solutions. Illustrative numerical results of our investigations on four regular types of zoom systems, as classified by Tanaka, are presented.