PURPOSE: Research shows wide variation in macular pigment density between individuals. As are other ocular pigments, this variation may be genetic. To test this hypothesis, the authors measured macular pigment density, serum carotenoid concentrations, and general dietary patterns in 10 pairs of identical twins. METHODS: Macular pigment was measured psychophysically by a 1 degree test stimulus. Foveal and parafoveal sensitivities to 460-nm and 530-nm light were compared. Determining the difference in log sensitivity to the 460-nm light for the fovea (where macular pigment is most dense) and the parafovea (where macular pigment is optically immeasurable), after normalizing with respect to 530 nm, yields a measurement of the optical density of macular pigment. Concentrations of carotenoids within the serum were measured using reverse-phase, high-performance liquid chromatography. Dietary patterns were determined using a food-frequency questionnaire. RESULTS: Statistically significant differences in macular pigment optical density were found for 5 of the 10 twin pairs. For these five pairs, differences in macular pigment density were moderately related to differences in the intake of dietary fat, iron, linoleic and oleic acid, fiber, and total calories (P < 0.10, individually; P < 0.05, for an equally weighted composite of these variables). There was no significant relationship, however, found between macular pigment density and carotenoids in the blood and diet. CONCLUSIONS: Given the putative protective role of macular pigment, variations in macular pigment density may have clinical significance. The conclusion that macular pigment is not completely determined genetically allows the possibility that macular pigment density may be modified for the protective purposes. The current data suggest that dietary fat, iron, and fiber may influence macular pigment levels (perhaps through their influence on carotenoid metabolism). These data suggest that the eventual deposition of macular pigment in the retina is complex and probably is influenced by a number of variables.