Many activities of magnesium have justified randomized controlled trials of its role in acute myocardial infarction (AMI), which have shown reduction of short term mortality by 25% to over 50%. The Fourth International Study of Infarct Survival (ISIS-4) megastudy failed to confirm these findings, and, based on analysis of pooled findings, it was concluded that magnesium has no place in treatment of AMI. The fixed effects statistical model employed in ISIS-4 for evaluation of pooled data is inappropriate because the studies were not homogeneous. Among the differences between the earlier studies and the megatrial, the most significant was the time at which magnesium infusions were started relative to the time of reperfusion. Animal studies have shown that magnesium is protective only if present before or at the time of reperfusion. Unlike in earlier trials, in which magnesium infusions were started soon after the ischemic event or simultaneously with a lytic agent, in ISIS-4 magnesium treatment was withheld until after iatrogenic or spontaneous reperfusion occurred. This can explain poor therapeutic results in ISIS-4, but not the hypotension and bradycardia encountered in a minority of patients in that study. Dosage difference alone cannot explain this, even though the amounts given in the small studies were 40% to 25% less than that in ISIS-4, because the dose used in the Second Leicester Intravenous Magnesium Intervention Trial (LIMIT-2) was only slightly lower than that used in ISIS-4. Administration of high dose magnesium with an angiotensin-converting enzyme inhibitor (which spares magnesium) or the vasodilating oral nitrate in arms of ISIS-4 may have contributed to adverse effects of hypermagnesemia. Also, the very low mortality rate of controls in ISIS-4 suggests that the patients may have been at relatively low risk, and it is in high risk patients that magnesium has been shown to be most effective. A large scale study of magnesium in such patients is being started.