BACKGROUND: Several social, demographic and physical factors have been shown to be associated with depression in later life, but results have been inconsistent. We aimed to assess factors associated with depression in old age, using data from the MRC trial of assessment and management of older people in the community. METHOD: Analysis of cross-sectional data. Depression was measured with a threshold of <6/6+ on the GDS-15. Independent associations with social, demographic, physical and social network variables were assessed by logistic regression. RESULTS: In a fully adjusted analysis, odds ratios (OR) for depression were greater in older people, (adjusted OR for those aged 80-84 years = 1.1, 85-90 years = 1.5 and 90+ years = 1.8), those in rented (OR:1.3) or sheltered/residential accommodation (OR:1.5), and those widowed, divorced or separated (OR:1.2). Life events, (OR:1.4), smoking (OR:1.6), having two or more physical illnesses (OR:1.6) or no confiding relationship (OR:3.4) were also significantly associated with depression. Higher alcohol consumption was not predictive. Female sex and living alone were associated with depression in a crude analysis, but not after full adjustment for confounding. CONCLUSIONS: The importance of these results lies in the large size and representative nature of the sample. In contrast to some previous reports, increasing age was associated with increasing risk of depression, but sex, living alone and alcohol were not associated. Social isolation was more important than living alone per se. Other associations largely concurred with previous work.