Anorexia nervosa (AN), bulimia nervosa (BN) and binge-eating disorder (BED) are characterized by abnormal eating behaviors often resulting in dramatic physical consequences for the patients. The etiology of eating disorders (EDs) is currently unknown; however, a strong genetic contribution is likely to be involved. To date, the majority of genetic studies have focused on candidate genes, and polymorphic variants of genes coding for substances likely to be involved in the etiopathogenesis of EDs have been assessed for association with AN, BN, BED and/or ED-related phenotypic traits. Results have been generally inconsistent and cannot be considered conclusive because of several methodological flaws and differences, such as small sample sizes, ethnic heterogeneity of studied populations, lack of statistical correction for multiple testing, adoption of different diagnostic criteria and population stratification. Although, at present, no convincing evidence for associations of candidate genes with EDs has been provided, the 5-HT(2A) receptor gene and the BDNF gene seem to be promising candidates for genetic influences on AN, since polymorphic variants of these genes have been found quite consistently, although not specifically, linked to AN restricting subtype in large sample studies. Moreover, pharmacogenetic investigations have suggested a possible role of some gene polymorphisms in predicting the response to treatment with selective serotonin reuptake inhibitors in BN, but results are still preliminary. The heterogeneity of ED phenotypes is believed to represent the most relevant variable responsible for contradictory and not conclusive results. Future studies should focus on more homogeneous subgroups, either relying on specific ED traits or identifying endophenotypes. This will be useful also for prevention and treatment of EDs.