To measure the prevalence and associated factors of undetected depression in institutionalized older people. Epidemiologic cross-sectional study in nursing homes and residential facilities. A stratified cluster sample of residents 65 years of age and older living in institutions of Madrid (Spain). Residents were considered to be depressed if they met at least 1 of the following 3 criteria: 10-item Geriatric Depression Scale score of 4 or higher, physician's diagnosis, or antidepressant use. Prevalence of undetected depression was defined as the proportion of depressed residents without documented diagnosis or treatment. A total of 255 of 579 residents had depression (weighted prevalence 46.1%, 95% confidence interval CI 41.0%-51.3%) and 108 depressed residents were undetected (undetection prevalence 41.5%, 95% CI 33.2%-50.2%). Undetection was lower in younger residents, private versus public facilities (sex-, age-, and size-adjusted prevalence ratio PR 0.59, 95% CI 0.37-0.94), and larger facilities (sex-, age-, and ownership-adjusted PR 0.94 per 50-bed increase, 95% CI 0.88-1.00). Undetected depression was higher in residents with poor self-rated health (sex- and age-adjusted PR 1.83, 95% CI 1.24-2.73), whereas the opposite came about for physician-rated health (PR 0.65, 95% CI 0.44-0.95). Undetection decreased 11% (95% CI 4%-17%) per 1-medication increase, and it was lower in patients with Alzheimer disease, anxiety, and arrhythmia. Number of medications and self-rated health were the main determinants of undetected depression. Physician-rated health, facility characteristics (size and ownership), and some diseases could also be considered.