Objectives We examined the associations between treatment attitudes and beliefs with race–gender differences in antidepressant adherence. Methods Subjects (n = 186) were African‐American and White subjects aged ≥60 years, diagnosed with clinically significant depression, and had a new outpatient primary care recommendation for antidepressant treatment. Antidepressant adherence was assessed using the Brief Medication Questionnaire. Attitudes and beliefs were assessed using the Patients Attitudes Toward and Ratings of Care for Depression, two items rating perceived medication importance, and a modified version of the Stigma Scale for Receiving Psychological Help. Results African‐American men and women had significantly greater concerns about antidepressants and significantly less understanding about treatment than White women. African‐American men had significantly more negative attitudes toward healthcare providers than African‐American and White women. African‐American women were more likely than White men and women to endorse a medication other than their antidepressant as most important. Whereas some race–gender differences were found in personal spirituality, no group differences were found in perceived stigma. In a logistic regression model adjusted for key baseline variables, White women were significantly more adherent to antidepressants than African‐American women (OR = 3.05, 95% CI = 1.06–8.81). Fewer concerns about antidepressants and indicating the antidepressant as the most important medication were both significantly associated with adherence. After including either of these two variables, the adherence difference between White women and African‐American women was no longer significant (OR = 2.56, 95% CI = 0.84–7.80). Conclusions Concerns about antidepressants and the importance of antidepressant medication are associated with adherence and are potentially modifiable through improved patient–provider communication, psycho‐education, and therapeutic interventions. Copyright © 2013 John Wiley & Sons, Ltd.