Restenosis after percutaneous transluminal coronary angioplasty (PTCA) is strongly associated with incomplete initial dilatation. To determine if oversized PTCA balloons would reduce the restenosis rate without increasing the risk of arterial dissection and acute complications, we prospectively randomized 336 patients to receive either smaller or larger balloons. Thirty-four percent of patients had multivessel disease and 18% had multisite dilatation. One hundred sixty-nine patients were randomized to PTCA with a larger balloon and 167 to PTCA with a smaller balloon. Balloon:artery diameter ratios were 1.13 +/- 0.14 in the larger group and 0.93 +/- 0.12 in the smaller group (p less than 0.001). The trial was halted as clinically important differences in acute complications emerged. Emergency bypass graft surgery, usually for the treatment of arterial dissection, was required in 7.1% of patients in the larger balloon group and 3.6% of patients in the smaller balloon group (p = 0.15). Myocardial infarction (Q wave and non-Q wave) complicated 7.7% of procedures in which large balloons were assigned and 3.0% of procedures in which small balloons were assigned (p = 0.056). There were no deaths in either group. The incidence of bypass surgery was 1.7% when the balloon:artery ratio was less than 0.9, 3.1% when the ratio was 0.9-1.1, and 7.8% when it was greater than 1.1. Stepwise logistic regression analysis demonstrated that larger balloon assignment, multiple lesion dilatation, and multivessel coronary artery disease were independent predictors of emergency surgery. Angiographic restudy rates were 50% in the larger group and 60% in the smaller group (p = NS).(ABSTRACT TRUNCATED AT 250 WORDS)