Between mid-1981 and Jan 1, 1985, 16 027 patients entering 245 coronary care units at a mean of 5.0 h after the onset of suspected acute myocardial infarction were randomised either to a control group or to a group receiving atenolol (5-10 mg iv immediately, followed by 100 mg/day orally for 7 days). Vascular mortality during the treatment period (days 0-7) was significantly lower (2p less than 0.4) in the treated group, 313/8037 (3.89%) versus 365/7990 (4.57%), but this 15% difference has wide 95% confidence limits (from about zero to about a quarter). No subgroups were identified in which the proportional difference in days 0-7 was clearly better, or clearly worse, than 15%. After the treatment period, there was only a slight further divergence (691 vs 703 additional vascular deaths by Jan 1, 1985). Thus, overall vascular mortality was significantly lower in the atenolol group at one year (life-table estimates: 10.7% atenolol vs 12.0% control; 2p less than 0.01) but not at Jan 1, 1985 (crude percentages: 12.5% vs 13.4%; 2p less than 0.07). However, atenolol patients were more likely than controls to be discharged on beta-blockers, which can account for much of the additional difference in vascular mortality after day 7. Immediate beta-blockade increased the extent of inotropic drug use (5.0% vs 3.4%, 2p less than 0.0001), chiefly on days 0-1, but despite this most of the improvement in vascular mortality was seen during days 0-1 (121 vs 171 deaths). Treatment did not appear to decrease the number in whom cardiac enzymes rose to above twice the local upper limit of normal. Slightly fewer non-fatal cardiac arrests (189 vs 198) and reinfarctions (148 vs 161) were recorded in the atenolol group, neither difference being significant. Systematic review of fatal and of non-fatal events in ISIS-1 and in all other randomised trials of iv beta-blockade reinforces the suggestion that treatment reduces mortality in the first week by about 15%, but with a rather less extreme effect in days 0-1 than was observed in ISIS-1 alone. It also provides highly significant (2p less than 0.0002) evidence of an effect on the combined end-point of death, arrest, or reinfarction, suggesting that treatment of about 200 patients would lead to the avoidance of 1 reinfarction, 1 arrest, and 1 death during days 0-7. ISIS-1 suggests these early gains will persist.