BACKGROUND The objective of this study was to examine how the effect of depressive symptoms on cognitive function is modified by church attendance. METHODS We used a sample of 2759 older Mexican Americans. Cognitive function was assessed using the Mini-Mental State Examination (MMSE) at baseline, 2, 5, 7, and 11 years of follow-up. Church attendance was dichotomized as frequent attendance (e.g., going to church at least once a month) versus infrequent attendance (e.g., never or several times a year). Depressive symptoms were assessed by the Center for Epidemiologic Studies Depression Scale (CES-D; score >or=16 vs <16). General linear mixed models with time-dependent covariates were used to explore cognitive change at follow-up. RESULTS In unadjusted models, infrequent church attendees had a greater decline in MMSE scores (drop of 0.151 points more each year, standard error [SE] = 0.02, p <.001) compared to frequent church attendees; participants having CES-D scores >or=16 also had greater declines in MMSE scores (drop of 0.132 points more each year, SE = 0.03, p <.001) compared to participants with CES-D score <16 at follow-up. In fully adjusted models, a significant Church attendance x CES-D x Time interaction (p =.001) indicated that, among participants with CES-D scores >or=16, infrequent church attendees had greater decline in MMSE scores (drop of 0.236 points more each year, SE = 0.05, p <.001) compared to frequent church attendees at follow-up. CONCLUSIONS Church attendance appears to be beneficial for maintaining cognitive function of older persons. Church attendance moderates the impact of clinically relevant depressive symptoms on subsequent cognitive function.