408 elderly primary care attenders were screened using the Geriatric Depression Scale (GDS) and the 30-item General Health Questionnaire (GHQ). 36% were identified as 'cases' on the GDS and 32% on the GHQ. The 'cases' and a random sample of 'non-cases' were interviewed using the Geriatric Mental State Examination (GMS), the Bedford College Life Events and Difficulties Interview (LEDS), the National Adult Reading Test (NART), and systematic inquiry concerning physical health. Sensitivity and specificity of the GDS were 85% and 68%, and of the GHQ 77% and 67%. General practitioner identification of cases showed a sensitivity of 78% and a specificity of 60% against the GMS. Depression was significantly associated with life events, chronic difficulties, poor physical health and current lack of a confiding relationship. Subjective complaints of depression were associated with low premorbid intelligence and lifelong lack of a confiding relationship. The availability of a confiding relationship appears to have a protective effect against depressive illness associated with life events and chronic difficulties.