Assessment of right ventricular volume and function is important in most congenital heart diseases before and after corrective or palliative surgery. Since transthoracic echocardiography is often substituted by transesophageal echocardiography in the perioperative setting, it is useful to compare transesophageal echocardiography with transthoracic echocardiography as performed preoperatively. We compared right ventricular volumes as calculated using these two methods from a four-chamber view in 21 children and adults with atrial septal defect. For right ventricular end-diastolic volumes of less than 70 ml, and end-systolic volumes of less than 40 ml, a close correlation was found between the techniques (r=0.99 and r=0.91, respectively), with a small degree of underestimation by transesophageal echocardio­graphy. For values larger than 70 nil and 40 ml, respectively, correlation decreased (r=0.41 for end-diastolic volumes and r=0.48 for end-systolic volumes) and underestimation of volume by transesophageal echocardiography increased. Underestimation of right ventricular end-diastolic volumes increased with increasing body surface area (r=0.74), and with progressive right ventricular enlargement (r=0.63). In patients with a body surface area of more than 1m2, the largest end-diastolic right ventricular length determined by transthoracic echocardiography was significantly longer than that derived by transesophageal echocardiography (p<0.001), whereas in smaller patients there was no significant difference between the two methods (p>0.1). If right ventricular volumes determined by transthoracic echocardiography using a four-chamber view are substituted by those obtained with transesophageal echocardiography in serial haemodynamic evaluation of patients with atrial septal defect, different correlation equations and, consequently, a different degree of underestimation by transesophageal echocardiography must be considered for large and small volumes. This increasing underestimation of larger right ventricular volumes seems to be based on foreshortening of the long cross-sectional axis of the right ventricle as seen in the transesophageal four-chamber view.