A core challenge in the genetic analysis of major depressive disorder (MDD) is how to recruit large numbers of stringently diagnosed cases with sufficient information to explore the interplay between genetic and environmental risk factors and evaluate genetic influences on putative MD subtypes and key clinical features. Currently, most genome-wide association studies of MDD rely on self-administered questionnaires or electronic health records, both of which are limited in diagnostic accuracy and introduce systematic, heritable biases that confound the interpretation of genetic analyses. Here, we describe how to address this problem through a combination of targeted ascertainment and in-depth phenotyping by clinical interview. We increase the homogeneity of the sample, reducing the number of cases needed to detect genetic signals, by recruiting only women with recurrent depressive episodes, ascertained through hospitals. Structured interviews capture detailed information on the known and putative risk factors for the disorder. We trained 347 interviewers working at 47 participating hospitals across South Korea and recruited 5704 cases and 4995 screened controls over 4 years toward a total target sample of 10,000 cases and 10,000 controls. We met and overcame a series of logistic challenges, including restrictions due to COVID-19 and an ongoing medical crisis. We confirmed that our cases have recurrent, severe MDD and are suitable to explore the causes of recurrent episodes of disturbances of sleep and appetite, suicidality, guilty ruminations, anhedonia, and low mood. Our study design provides deeply phenotyped cases and screened controls at scale and can be adapted for deployment in other countries to yield cohorts for the genetic analysis of MDD.