Background. The high risk of suicide in bipolar disorder is well recognized, but may have been overestimated. There is conflicting evidence about deaths from other causes and little known about risk factors for suicide. We aimed to estimate suicide and mortality rates in a cohort of bipolar patients and to identify risk factors for suicide.

Method. All patients who presented for the first time with a DSM-IV diagnosis of bipolar I disorder in a defined area of southeast London over a 35-year period (1965–1999) were identified. Mortality rates were compared with those of the 1991 England and Wales population, indirectly standardized for age and gender. Univariate and multivariate analyses were used to test potential risk factors for suicide.

Results. Of the 239 patients in the cohort, 235 (98·3%) were traced. Forty-two died during the 4422 person-years of follow-up, eight from suicide. The standardized mortality ratio (SMR) for suicide was 9·77 [95% confidence interval (CI) 4·22–19·24], which, although significantly elevated compared to the general population, represented a lower case fatality than expected from previous literature. Deaths from all other causes were not excessive for the age groups studied in this cohort. Alcohol abuse [hazard ratio (HR) 6·81, 95% CI 1·69–27·36, p=0·007] and deterioration from pre-morbid level of functioning up to a year after onset (HR 5·20, 95% CI 1·24–21·89, p=0·024) were associated with increased risk of suicide.

Conclusions. Suicide is significantly increased in unselected bipolar patients but actual case fatality is not as high as previously claimed. A history of alcohol abuse and deterioration in function predict suicide in bipolar disorder.