Reproduction and ewe productivity are complex composite traits that are influenced by several component traits. Genetic improvement by selection for an individual component trait may not always be advantageous because adverse or neutral genetic relationships can exist among the component traits. Selection for an overall composite trait of ewe productivity, defined as litter weight weaned per ewe joined, can result in a balanced biological composite trait with favourable responses in component traits including fertility, number of lambs born, lamb survival, lactation and lamb growth. Selection for litter weight weaned may also overcome adverse genetic relationships among reproductive component traits. This review examines the genetic responses of selecting directly for litter weight weaned, compared with selection responses for one or more of its component traits. It is concluded that selection to improve reproductive efficiency and ewe productivity, under most production and environmental systems, would benefit from selection for a composite trait such as litter weight weaned, rather than for a single component trait.