This article reviews the histopathologic and psychophysical evidence for selective losses to specific subpopulations of optic nerve fibers in glaucoma. Based on this evaluation, the degree to which these losses are selective is drawn into question, and a distinction is drawn between selective tests and selective losses. An alternative hypothesis for early detection of functional losses in glaucoma, the reduced redundancy hypothesis, is presented. This concept takes into account the redundancy or sampling characteristics of specific subpopulations of optic nerve fibers, as well as the relative amount of glaucomatous loss incurred by each optic nerve fiber subpopulation. An example is presented in which an undersampled subpopulation of optic nerve fibers with minimal redundancy is better able to reveal early losses, even though there are greater amounts of relative loss for other optic nerve fiber subpopulations. The design of psychophysical tests for early detection of functional losses in glaucoma should take into account both the relative amounts of loss for various subpopulations of optic nerve fibers and their inherent redundancy or sampling properties.