OBJECTIVE: To examine the longitudinal relationship between depression outcomes and subsequent functional disability (FD) in the non-institutionalized elderly. METHODS: Secondary data analysis was performed using the population-based Assets and Health Dynamics of the Oldest Old (AHEAD) cohort (age =70 years). Depression was considered present if four or more depressive symptoms were reported on the modified Center for Epidemiological Studies-Depression Scale (CES-D). CES-D scores from baseline, two-year and five-year follow-up allowed the characterization of seven distinct patterns of depression (plus never depressed): remittent, endogenous, emergent, remitting persistent, recurrent, emerging persistent and persistent. FD was operationalized as the ability to perform six activities of daily living (ADL) and five instrumental ADL. The subsequent impact of depression patterns on FD scores (at two, five and seven-year follow-up) over time was analyzed using mixed-effect regression models. RESULTS: Of the 8222 initial respondents, 57% were considered ineligible. Among the remaining 3476 respondents, half were never depressed. Pattern-based FD mean (SD) scores were: remittent 1.4 (2.4) (n=166); endogenous, 1.4 (2.2) (n=136); recurrent, 2.03 (2.6) (n=63); emergent 1.2 (2.2) (n=332); remittent persistent 1.9 (2.5) (n=64); recurrent 2.0 (2.6) (n=103); emerging persistent 2.1 (2.6) (n=108) and persistent 2.1 (2.5) (n=144). After adjusting for age, gender, and comorbidity, and baseline FD, all patterns had significantly more FD than those never depressed, with the exception of the remittent pattern. Compared to an emergent pattern, emerging persistent (difference=0.69 (0.19), p<0.001) and persistent pattern (difference=0.49 (0.15), p<0.001) had higher mean FD over time. CONCLUSIONS: The elderly with depression have more FD compared to those never depressed, but FD can improve with the remission of depression. Important differences in FD scores between depression patterns were observed (i.e. effect sizes >0.5), providing health outcomes-based support for a pattern-based classification of longitudinal depression.