Genetic predisposition may contribute to the differences in drug-specific, class-specific or antidepressant-wide treatment resistance. Clinical studies with the genetic data are often limited in sample sizes. Drug response obtained from self-reports may offer an alternative approach to conduct a study with much larger sample size. Using the phenotype data collected from 23andMe 'Antidepressant Efficacy and Side Effects' survey and genotype data from 23andMe's research participants, we conducted genome-wide association study (GWAS) on subjects of European ancestry using four groups of phenotypes (a) non-treatment-resistant depression (n=7795) vs treatment-resistant depression (TRD, n=1311), (b) selective serotonin reuptake inhibitors (SSRI) responders (n=6348) vs non-responders (n=3340), (c) citalopram/escitalopram responders (n=2963) vs non-responders (n=2005), and (d) norepinephrine-dopamine reuptake inhibitor (NDRI, bupropion) responders (n=2675) vs non-responders (n=1861). Each of these subgroups was also compared with controls (n ~ 190 000). The most significant association was from bupropion responders vs non-responders analysis. Variant rs1908557 (P=2.6 × 10(-8), OR=1.35) passed the conventional genome-wide significance threshold (P=5 × 10(-8)) and was located within the intron of human spliced expressed sequence tags in chromosome 4. Gene sets associated with long-term depression, circadian rhythm and vascular endothelial growth factor (VEGF) pathway were enriched in the bupropion analysis. No single-nucleotide polymorphism passed genome-wide significance threshold in other analyses. The heritability estimates for each response group compared with controls were between 0.15 and 0.25, consistent with the known heritability for major depressive disorder.