A prospective twin study on age-related macular degeneration (AMD) recruited 83 monozygotic pairs, 28 dizygotic pairs, and one triplet set from 1986 through 1993. Zygosity was determined by genetic testing of red cell markers, HLA antigens, or specific DNA loci. There were no twin pairs in which I collected data on only one twin. To decrease ascertainment bias, after 1991 the recruitment notice did not mention AMD, and I did not ask about a history of eye disease before the eye examination. Because of this, twin pairs recruited from 1986 through 1991 were statistically analyzed separately from those after January 1, 1992. From 1986 through 1991, 23 twin pairs were recruited; 11 monozygotic and 2 dizygotic pairs had nonAMD retinal changes or no retinal abnormalities, 9 monozygotic pairs with AMD were all concordant, and 1 dizygotic pair was discordant for basal laminar drusen. The concordance rate of AMD did not differ significantly between monozygotic and dizygotic twin pairs (P = .10) for 1986 through 1991. In 1992 and 1993, 88 twin pairs and one triplet set were recruited; 49 monozygotic and 19 dizygotic pairs had nonAMD retinal changes or no retinal abnormalities, 14 monozygotic pairs with AMD were all concordant, and 2 of 7 dizygotic pairs were concordant for AMD. The nonidentical triplets (1 with and 2 without AMD) were categorized as one of the discordant dizygotic pairs in the statistical evaluation. In nontwin age-matched (within 2 or 5 years of age) or age- and sex-matched sibling pairs the concordance rate of AMD ranged from 16% to 25%. The concordance rate of AMD was significantly higher in monozygotic than in dizygotic twins (P = .001) for 1992 and 1993. The concordance rate was higher for monozygotic twin pairs recruited in 1992 and 1993 than in any of the four subsets of nontwin age-method or age- and sex-matched sibling pairs (P < .0001). Overall, from 1986 through 1993, 23 of 23 monozygotic and 2 of 8 dizygotic twin pairs were concordant for AMD; this included the one dizygotic pair which was discordant for basal laminar drusen. The data of this study strongly suggest a genetic predisposition to AMD.