Over the past 20 years at least 11 randomised trials of the prevention with diuretics of pre-eclampsia and its sequelae have been undertaken. Nine of these were reviewed. Reliable data from the remaining two were not available. The nine reviewed had investigated a total of nearly 7000 people. Significant evidence of prevention of "pre-eclampsia" was overwhelming, even when oedema was not included as a diagnostic criterion. But as the definitions of pre-eclampsia that had been used depended heavily on increases in blood pressure this evidence may simply have reflected the well known ability of diuretics to reduce blood pressure. When the data on perinatal death were reviewed a little difference was seen in postnatal survival. The incidence of stillbirths was reduced by about one third with treatment, but, perhaps owing to small numbers (only 37 stillbirths), the difference was not significant. Thus these randomised trials failed to provide reliable evidence of either the presence or the absence of any worthwhile effects of treatment with diuretics on perinatal mortality. The implications of this for current and future trials of beta blockers and other agents in the prevention of pre-eclampsia and its sequelae are that extremely large, ultra simple randomised trials are needed, of a size sufficient to permit direct assessment of the effects of treatment not on pre-eclampsia but on perinatal mortality itself. This may require the study of tens of thousands of pregnancies.