Objectives To characterize the health-care utilization and economic burden associated with depression in Manitoba, Canada. Methods Patient-level data were retrieved from the Manitoba Centre for Health Policy administrative, clinical, and laboratory databases for the study period of January 1, 1996, through December 31, 2016. Patients were assigned to the depression cohort based on diagnoses recorded in hospitalizations and outpatient physician claims, as well as antidepressant prescription drug claims. A comparison cohort of nondepressed subjects, matched with replacement for age, gender, place of residence (urban vs. rural), and index date, was created. Demographics, comorbidities, intentional self-harm, mortality, health-care utilization, prescription drug utilization, and costs of health-care utilization and social services were compared between depressed patients and matched nondepressed patients, and incidence rate ratios and hazard ratios were reported. Results There were 190,065 patients in the depression cohort and 378,177 patients in the nondepression cohort. Comorbidities were 43% more prevalent among depressed patients. Intentional self-harm, all-cause mortality, and suicide mortality were higher among patients with depression than the nondepression cohort. Health-care utilization—including hospitalizations, physician visits, physician-provided psychotherapy, and prescription drugs—was higher in the depression than the nondepression cohort. Mean health-care utilization costs were 3.5 times higher among depressed patients than nondepressed patients ($10,064 and $2,832, respectively). Similarly, mean social services costs were 3 times higher ($1,522 and $510, respectively). Overall, depression adds a total average cost of $8,244 ( SD = $40,542) per person per year. Conclusions Depression contributes significantly to health burden and per patient costs in Manitoba, Canada. Extrapolation of the results to the entire Canadian health-care system projects an excess of $12 billion annually in health system spending.