Background: Autoantibodies have been implicated in the etiologic pathway of depressive disorders. Here, we determine the association between the presence of a panel of autoantibodies at baseline and change in depression symptom score over 5-year follow-up in a cohort of healthy elderly Australians. Methods: Serum samples from 2049 randomly selected subjects enrolled in the Hunter Community Study (HCS) aged 55-85 years were assayed for a range of autoimmune markers (anti-nuclear autoantibodies, extractable nuclear antigen autoantibodies, anti-neutrophil cytoplasmic autoantibodies, thyroid peroxidase autoantibodies, tissue transglutaminase autoantibodies, anti-cardiolipin autoantibodies, rheumatoid factor and cyclic citrullinated peptide autoantibodies) at baseline. Depression symptom score was assessed using the Centre for Epidemiological Study (CES-D) scale at baseline and 5 years later. Results: Autoantibody prevalence varied amongst our sample with ANA being the most prevalent; positive in 16% and borderline in 36% of study population. No evidence for a relationship was found between change in CES-D score over time and any autoimmune marker. Statins and high cholesterol were significantly associated with change in CES-D score over time in univariate analysis; however, these were probably confounded since they failed to remain significant following multivariable analysis. Conclusions: Autoantibodies were not associated with change in CES-D score over time. These findings point to an absence of autoimmune mechanisms in the general population or in moderate cases of depression.