Twenty-five patients with transposition of the great arteries underwent primary arterial switch repair between July 1985 to October 1986. Twenty infants had intact ventricular septum or small ventricular septal defect while 5 infants also had moderate or large ventricular septal defects. Median age of switch repair was 2.9 weeks in those with small ventricular septal defect or intact septum and 7.6 weeks in those with significant defects. Vectorcardiograms were performed weekly preoperatively and then up to 4 to 6 months postoperatively. Cardiac catheterisation was performed preoperatively. Correlation between a left-to-right ventricular pressure ratio of more or less than 0.7 was made with a vectorcardiogram taken near cardiac catheterization. Preoperative vectorcardiograms in patients with essentially intact ventricular septum showed normal maximal spatial voltages in 60 and 65% (left and right) although 70% had clockwise horizontal loops. There was poor correlation between maximal spatial voltages and loops of the vectorcardiogram with the left-to-right ventricular pressure ratio at catheterization. Three of 5 patients with large ventricular septal defects had decreased left maximal spatial voltages or clockwise loops preoperatively. This suggests low left ventricular pressure ratio but at catheterization all had elevated pressure ratios. Despite decreased left maximal spatial voltages or clockwise horizontal loops in many patients in both groups, all survived the arterial switch procedure. Postoperatively, the vectorcardiogram normalized in those with essentially intact septum within a mean of 4 weeks but took longer in the others. The vectorcardiogram is unreliable in assessing suitability for the arterial switch procedure in the first 3 months of life.(ABSTRACT TRUNCATED AT 250 WORDS)