IMPORTANCE: Antidepressant treatment remains a trial-and-error process: one-third of people with major depressive disorder (MDD) report inefficacy of first-line medications. Predictors of prescription patterns are needed to improve prescribing precision. OBJECTIVE: To investigate phenotypic and genetic heterogeneity of MDD subgroups defined by antidepressant prescription patterns. DESIGN, SETTING, AND PARTICIPANTS: This was a retrospective cohort study of Australian Genetics of Depression Study (2017-2018) adult participants (aged ≥18 years) with lifetime MDD who filled 1 or more prescriptions of the 10 most commonly used antidepressants across 4.5 years (2013-2017). Data were analyzed from August 2024 to October 2025. EXPOSURES: Treatment complexity was assessed as number of different antidepressant classes in prescriptions filled in 4.5 years. Sustained-use 360 groups were defined as 360 or more cumulative days (in 4.5 years) of a single antidepressant. Participants with genome-wide genotypes were contrasted across mutually exclusive sustained-use 360 groups. MAIN OUTCOMES AND MEASURES: Associations of 44 self-reported phenotypes and polygenic scores (PGSs) for 15 traits with sustained-use 360 subgroups. Genome-wide association studies (GWASs) were conducted for selective serotonin reuptake inhibitor (SSRI) or SSRI/serotonin-norepinephrine reuptake inhibitor sustained use contrasted to other participants. RESULTS: Of 12 074 participants (9041 [75%] female with a mean [SD] age of 41.8 [14.6] years; 3022 [25%] male with a mean [SD] age of 47.7 [14.6] years) with 1 or more prescriptions and lifetime MDD, 8898 had genotyping data. High treatment complexity was significantly associated with 37 of 44 self-reported phenotypes (eg, higher rates of smoking, recurrent MDD, suicidal ideation, chronic pain, and circadian and atypical depression subtypes) and higher PGSs for psychiatric traits (MDD PGS: β, 0.04; 95% CI, 0.03-0.06; P = 1.2 × 10-8; ADHD PGS: β, 0.03; 95% CI, 0.02-0.05; P = 2.1 × 10-5 ; bipolar disorder PGS: β, 0.03; 95% CI, 0.01-0.04. P = 1.2 × 10-4 ; neuroticism PGS: β, 0.02; 95% CI, 0.01-0.04; P = 1.3 × 10-3). A total of 5453 (61%) met criteria for an exclusive antidepressant sustained-use 360 group. These groups had distinct phenotypic profiles, including associations with body mass index, suicidal ideation, and co-occurring conditions. GWASs identified novel loci, including an immune-related gene SLAMF3/LY9, for which single-nucleotide variant rs4656934 was associated with reduced odds of sustained SSRI use (G allele; odds ratio, 0.81; 95% CI, 0.75-0.87; P = 3.5 × 10-8). CONCLUSIONS AND RELEVANCE: This study found that phenotypic factors were associated with sustained antidepressant use and treatment complexity. PGSs for traits studied were associated with treatment complexity but showed little association with sustained-use 360 groups. These findings support further research to guide treatment selection and to identify patients at risk of difficult-to-treat depression, informing precision psychiatry and early intervention in MDD.