Background Alcohol dehydrogenase 1B and 1C (ADH1B and ADH1C) variants have been robustly associated with alcohol phenotypes in East Asian populations, but less so in non‐Asian populations where prevalence of the most protective ADH1B allele is low (generally <5%). Further, the joint effects of ADH1B and ADH1C on alcohol phenotypes have been unclear. Therefore, we tested the independent and joint effects of ADH1B and ADH1C on alcohol phenotypes in an Israeli sample, with higher prevalence of the most protective ADH1B allele than other non‐Asian populations. Methods A structured interview assessed lifetime drinking and alcohol use disorders (AUDs) in adult Israeli household residents. Four single nucleotide polymorphisms (SNPs) were genotyped: ADH1B (rs1229984, rs1229982, and rs1159918) and ADH1C (rs698). Regression analysis examined the association between alcohol phenotypes and each SNP (absence vs. presence of the protective allele) as well as rs698/rs1229984 diplotypes (also indicating absence or presence of protective alleles) in lifetime drinkers (n = 1,129). Results Lack of the ADH1B rs1229984 protective allele was significantly associated with consumption‐ and AUD‐related phenotypes (OR = 1.77 for AUD; OR = 1.83 for risk drinking), while lack of the ADH1C rs698 protective allele was significantly associated with AUD‐related phenotypes (OR = 2.32 for AUD). Diplotype analysis indicated that jointly ADH1B and ADH1C significantly influenced AUD‐related phenotypes. For example, among those without protective alleles for ADH1B or ADH1C, OR for AUD was 1.87 as compared to those without the protective allele for ADH1B only and was 3.16 as compared to those with protective alleles for both ADH1B and ADH1C. Conclusions This study adds support for the relationship of ADH1B and ADH1C and alcohol phenotypes in non‐Asians. Further, these findings help clarify the mixed results from previous studies by showing that ADH1B and ADH1C jointly effect AUDs, but not consumption. Studies of the association between alcohol phenotypes and either ADH1B or ADH1C alone may employ an oversimplified model, masking relevant information.