Non-adherence to antidepressant medication is a well-established factor contributing to treatment failure among patients with major depressive disorder. Addressing this issue is crucial not only for enhancing individual patient outcomes but also for alleviating the broader public health burden. The study aimed to assess antidepressant medication non-adherence and its determinants among patients with major depressive disorder at public hospital psychiatric clinics in Ethiopia. Between June 12, 2024, and November 13, 2024, a multicenter cross-sectional study was conducted at public hospital psychiatric clinics in Ethiopia. Antidepressant non-adherence was assessed using a self-reported tablet count tool with pharmacy refill records available in patients' charts at the time of the interview. The severity of adverse drug reactions (ADRs) was evaluated using the Antidepressant Side-Effect Checklist (ASEC), while the Naranjo ADR Probability Scale was employed to determine the likelihood of ADRs. Data analysis was performed using SPSS version 26.0. Frequencies and percentages, were used to describe the characteristics of study participants. For factors associated with antidepressant non-adherence, multivariate logistic regression was conducted. The association between explanatory variable and non-adherence was assessed using odds ratios (ORs) with 95% CIs. The prevalence of antidepressant medication non-adherence was 139 (32.9%). Female gender [AOR = 3.29, 95% CI (2.04, 5.31)], illiteracy [AOR = 2.17, 95% CI (1.35, 3.50)], unemployment [AOR = 3.40, 95% CI (2.15, 5.38)], treatment duration greater than 25 months [AOR = 1.89, 95% CI (1.05, 3.41)], and severe ADRs [AOR = 3.94, 95% CI (1.68, 9.23)] were significantly associated with Antidepressant medication non-adherence. Being female, illiterate, unemployed, having a treatment duration of more than 25 months, and experiencing severe adverse drug reactions were significantly associated with non-adherence. These findings highlight the need for targeted interventions to improve adherence among these high-risk groups.