OBJECTIVE Parenterally administered MgSO4 is neuroprotective in standard animal models of focal cerebral ischemia and in many other paradigms of brain injury. Previous small clinical trials in stroke patients have explored the safety and tolerability of different infusion regimens. This study was undertaken to optimize the regimen for a multicenter trial. METHODS Within 24 hours of the onset of clinically diagnosed stroke, patients were randomized to receive placebo or one of three intravenous MgSO4 infusions: a loading infusion of 8, 12, or 16 mmol, followed by 65 mmol over 24 hours. Cardiovascular parameters, serum magnesium concentrations, and blood glucose concentrations were determined. Outcome at 30 and 90 days was recorded. RESULTS Twenty-five patients were recruited and treated at a mean time of 20 hours after stroke. No tolerability problems were identified. No effects of magnesium on heart rate, blood pressure, or blood glucose were evident. Serum magnesium concentrations rose to target levels most rapidly in the highest loading infusion group and were maintained in all groups for at least 24 hours. CONCLUSIONS MgSO4 infusions that rapidly elevate the serum magnesium concentration to potentially therapeutic levels are well tolerated and have no major hemodynamic effects in patients with acute stroke. The 16-mmol loading infusion achieved target serum concentrations most rapidly and has been chosen for further trials.