Summary. The superposition eyes and simple eyes of many arthropods have apertures (A) with a diameter bigger than, or about the same size as, the focal length of the eye (f). That is, these eyes have low F-numbers (f/A). Many of the light rays focussed onto a photoreceptor will not be trapped by total internal reflection in the photoreceptor and will therefore pass through and be absorbed in photoreceptors other than that for which the light was intended. This spread of light in the retina leads to a broadening of the angular-sensitivity function and a consequent degrading of the image at the retinal level. A number of solutions to this problem are found in nature, with the most effective that of isolating the photoreceptors with a sheath of either light-absorbing pigment or reflecting tapetum. A ray-tracing model was used to assess the relative merits of the tapetal and pigment sheath designs in low F-number superposition eyes, and also to investigate the effect of changing the refractive index and absorption coefficient of the rhabdom. Which sheathing solution is best depends on the quality of the image on the retina, on the spacing of the rhabdoms in the retina and on the intensity of light normally experienced by the eye. In a retina with closely packed rhabdoms, the model predicts full sheathing to be the optimal solution if the image is well-focussed, partial sheathing if poorly focussed and no sheathing if moderately well-focussed. In a retina with rhabdoms spaced apart and a well-focussed image, the model predicts partial sheathing to be optimal. A pigment sheath is predicted to be useful in eyes which experience bright light and have no need for high sensitivity. A tapetal sheath is predicted to be useful at any intensity. A survey of arthropod eyes with low F-number supports the predictions of the model.