Background: Depression is common among older adults and is typically treated with antidepressants. Objective: To determine the non-adherence rates to antidepressants among older adults in primary care, based on non-initiation, suboptimal implementation or non-persistence. Methods: We selected all patients aged ≥60 years and diagnosed with depression in 2012, from the Netherlands Institute for Health Services Research (NIVEL) Primary Care Database. Non-initiation was defined as no dispensing within 14 days of the first prescription; suboptimal implementation, as fewer than 80% of the days covered by dispensed dosages; and non-persistence, as discontinuation within 294 days after first dispense. First, we determined the antidepressant non-initiation, suboptimal implementation and non-persistence rates. Second, we examined whether comorbidity and chronic drug use were associated with non-adherence by mixed-effects logistic regression (non-initiation or suboptimal implementation as dependent variables) and a clustered Cox regression (time to non-persistence). Results: Non-initiation, suboptimal implementation and non-persistence rates were 13.5%, 15.2% and 37.1%, respectively. As the number of chronically used drugs increased, the odds of suboptimal implementation (odds ratio, 0.89; 95% confidence interval, 0.83-0.95) and of non-persistence (hazard ratio, 0.87; 95% confidence interval, 0.82-0.92) reduced. Conclusions: Non-adherence to antidepressants is high among older patients with depression in primary care settings. Adherence is better when patients are accustomed to taking larger numbers of prescribed drugs, but this only provides partial explanation of the variance. GPs should be aware of the high rates of non-adherence. Emphasizing the importance of adhering to the optimal length of antidepressant therapy might be prudent first steps to improving adherence.