Mortality data from 3350 patients who attended the Glasgow Blood Pressure Clinic between 1968 and the end of 1982 were used to examine the hypothesis that lowering diastolic blood pressure (DBP) below 85 mmHg causes death from coronary heart disease (CHD). Analysis of 257 coronary deaths in quintiles of treated DBP showed a significantly nonlinear relation, with the lowest mortality from CHD occurring in the middle quintile (91-98 mmHg). This finding persisted after adjustment for risk at entry, and was independent of sex and pre-existing CHD. In contrast, the relations between treated systolic blood pressure and death from CHD, and treated DBP and death from stroke were linear. For 2355 patients who were untreated at referral there was no relation between the change in DBP during treatment and death from CHD. In our view, however, these findings do not necessarily support the hypothesis that lowering of DBP below 85 mmHg with treatment causes death from CHD. Evidence for this is indirect and inconsistent, and should not, at present, be used as a basis for any change in treatment practice.