This study examined whether the relationship between HAART medication adherence and antidepressant treatment varied with HAART regimen complexity. The analysis included 1,192 respondents to the HIV Cost and Service Utilization Study (HCSUS) who were taking HAART. Self-reported past-week HAART adherence, antidepressant use, current mental health status, and an aggregate measure of regimen complexity were used in the analysis. Regression models with interactions between antidepressant treatment of mental health problems, poor emotional well-being or depressive symptoms, and medication complexity were estimated to assess differential associations with adherence. Irrespective of antidepressant treatment, poor mental health status was negatively associated with HAART medication adherence. However, only untreated higher depressive symptoms were strongly associated with maladherence to HAART medication (OR = 0.72 at P < 0.05). Medication complexity was strongly associated with maladherence to HAART medication (OR = 0.96 with P < 0.05) in the model including emotional well-being; and weakly associated with maladherence (OR = 0.97 and P < 0.07) in the model including depressive and/or anxiety symptoms. However, as HAART medication complexity increased, adherence was higher among individuals with poorer mental health but using antidepressants compared to those with better mental health (OR = 1.09 with P < 0.05 in the model including emotional well-being; OR = 1.09 and P < 0.05 in the model including depressive and/or anxiety symptoms), and compared to those with poorer mental health but not using antidepressants (OR = 1.09, P = 0.08 in the model including emotional well-being, and OR = 1.12, P < 0.05 in model including depressive and/or anxiety symptoms). In conclusion, while individuals with poorer mental health generally have poor HAART adherence, their adherence improved with the use of antidepressants as the HAART complexity increased.