The production and prevention of calcium paradox injury in myocardium was studied in a canine model of cardiopulmonary bypass with multidose, moderately hypothermic, crystalloid cardioplegic solution. During 4 1/2 hours of global ischemia, three groups of six dogs each received one of three histidine-buffered cardioplegic solutions (500 ml initially and 250 ml every 30 minutes) at 27 degrees C. Group 1 cardioplegic solution was calcium free, group 2 solution contained a trace amount of calcium chloride (70 mumols /L), and group 3 cardioplegic solution was calcium free but contained diltiazem (150 micrograms/kg body weight). Left ventricular function measured as percent control of developed pressure revealed significantly greater (p less than 0.05) recovery in groups 2 and 3. Triphenyltetrazolium chloride staining showed 35% +/- 9% (mean +/- standard error) of heart mass necrosis in group 1 versus 0% and 0.5% +/- 0.4% in groups 2 and 3, respectively (p less than 0.001). Electron microscopy revealed ultrastructural changes characteristic of calcium paradox injury in group 1 myocardium. Calcium paradox injury was produced in an in vivo model of global myocardial ischemia and multidose cardioplegia despite moderate hypothermia and non-coronary collateral flow. The addition of either trace levels of calcium or diltiazem to the cardioplegic solution was effective in preventing this injury.