Mechanisms for the association between psychiatric disorder measured by the General Health Questionnaire and subjective and objective indices of physical illness are examined among 6900 male and 3414 female civil servants aged between 35 and 55 yr from the cross-sectional results of the Whitehall II Study. It was hypothesized that the association between physical illness and psychiatric disorder might be because psychiatric disorder was either secondary to the pain and disability of physical illness, or related to the somatic presentation of psychiatric disorder, or to common causes of both physical and psychiatric illness. Overall health status, and self-reported physical symptoms were strongly associated with psychiatric disorder. Angina in men and severe chest pain in men and women were significantly associated with psychiatric disorder, but electrocardiographic abnormalities were not. Established risk factors for physical illness: alcohol intake in women and men and smoking habit in men were related to psychiatric disorder. Fibrinogen was related to self reported physical symptoms but not psychiatric disorder. The association between psychiatric disorder and self-reported physical symptoms is most likely explained by somatisation and plaintive set. However, psychiatric disorder is likely to be secondary to the pain and disability of conditions such as angina.