58 050 patients entering 1086 hospitals up to 24 h (median 8 h) after the onset of suspected acute myocardial infarction (Ml) with no clear contraindications to the study treatments (in particular, no cardiogenic shock or persistent severe hypotension) were randomised in a "2 x 2 x 2 factorial" study. The treatment comparisons were: (i) 1 month of oral captopril (6·25 mg initial dose titrated up to 50 mg twice daily) versus matching placebo; (ii) 1 month of oral controlled-release mononitrate (30 mg initial dose titrated up to 60 mg once daily) versus matching placebo; and (iii) 24 h of intravenous magnesium sulphate (8 mmol initial bolus followed by 72 mmol) versus open control. There were no significant "interactions" between the effects of these three treatments, and the results for each are based on the randomised comparison of about 29 000 active versus 29 000 control allocated patients. Captopril There was a significant 7% (SD 3) proportional reduction in 5-week mortality (2088 [7·19%] captopril-allocated deaths vs 2231 [7·69%] placebo; 2p=0·02), which corresponds to an absolute difference of 4·9 SD 2·2 fewer deaths per 1000 patients treated for 1 month. The absolute benefits appeared to be larger (perhaps about 10 fewer deaths per 1000) in certain higher-risk groups, such as those presenting with a history of previous Ml or with heart failure. The survival advantage appeared to be maintained in the longer term (5·4 [SD 2·8] fewer deaths per 1000 at 12 months). Captopril was associated with an increase of 52 (SD 2) patients per 1000 in hypotension considered severe enough to require termination of study treatment, of 5 (SD 2) per 1000 in reported cardiogenic shock, and of 5 (SD 1) per 1000 in some degree of renal dysfunction. It produced no excess of deaths on days 0-1, even among patients with low blood pressure at entry. Mononitrate There was no significant reduction in 5-week mortality, either overall (2129 [7·34%] mononitrate-allocated deaths vs 2190 [7·54%] placebo) or in any subgroup examined (including those not receiving short-term non-study intravenous or oral nitrates at entry). Further follow-up did not indicate any later survival advantage. The only significant side-effect of the mononitrate regimen studied was an increase of 15 (SD 2) per 1000 in hypotension. Those allocated active treatment had somewhat fewer deaths on days 0-1, which is reassuring about the safety of using nitrates early in acute MI. Magnesium There was no significant reduction in 5-week mortality, either overall (2216 [7·64%] magnesium- allocated deaths vs 2103 [7·24%] control) or in any subgroup examined (including those treated early or late after symptom onset or in the presence or absence of fibrinolytic or antiplatelet therapies, or those at high risk of death). Further follow-up did not indicate any later survival advantage. In contrast with some previous small trials, there was a significant excess with magnesium of 12 (SD 3) per 1000 in heart failure and of 5 (SD 2) per 1000 in reported cardiogenic shock during or just after the infusion period. Magnesium was also associated with an increase of 11 (SD 2) per 1000 in hypotension considered severe enough to require termination of study treatment, of 3 (SD 0·6) per 1000 in bradycardia, and of 3 (SD 0·4) per 1000 in a cutaneous flushing or burning sensation (but assessment of magnesium involved open control). There was no evidence of a net adverse effect on mortality on days 0-1. Because of its size, ISIS-4 provides reliable evidence about the effects of adding each of these three treatments to established treatments for acute Ml. Intravenous magnesium was ineffective, and although oral nitrate therapy appeared safe it did not produce a clear reduction in 1-month mortality. Other trials have shown that starting long-term converting enzyme inhibitor (CEI) therapy in the weeks or months after Ml in patients with impaired ventricular function avoids about 2 deaths per 1000 patients per month of treatment. ISIS-4, GISSI-3, and smaller studies now collectively demonstrate that, for a wide range of patients without clear contraindications, CEI therapy started early in acute MI prevents about 5 deaths per 1000 in the first month (2p=0·006), with somewhat greater benefits in higher-risk patients. This benefit from 1 month of early CEI treatment seems to persist for at least the first year.