Both 2-D echocardiography and angiocardiography were performed preoperatively in 40 children with isolated ventricular septal defect (VSD). The diagnosis of VSD was confirmed and the size of VSD was measured during operation within on month of the study. There were 27 males and 13 females (age: 10mos-9 yrs). The VSD was classified into perimembranous type (24), supracristal type (14), and AV canal type (2) by surgery. The maximal VSD areas were calculated by 1/4 pi De2 (De = maximal defect diameter measured by 2D echo). The maximal shunt flow areas were measured by color Doppler echocardiography from frame by frame search of the video tape. The maximal shunt flow area was corrected by the body surface area to be the maximal shunt flow area index. A left ventriculogram (LVG) was performed from 4-chamber and left lateral views to measure the maximal diameter of the defect. The maximal defect area of VSD on angiocardiogram was also calculated by 1/4 pi Da2 (Da = maximal defect diameter measured by angiocardiography). The VSD areas measured by different methods were compared by the linear regression analysis. We found that: (1) there is good correlation of the maximal defect diameter and the area of ventricular septal defect by 2-DE and LVG (r = 0.80, p less than 0.0001 v.s. r = 0.78, p less than 0.0001). (2) the ventricular septal defect area by 2-DE had a better correlation with the surgical measurement than that by LVG (r = 0.93, p less than 0.0001, v.s. r = 0.80, p less than 0.0001). (3) the size of ventricular septal defect measured by two-dimensional echocardiography has a better correlation with the MSFAI than the LVG did (r = 0.81, p less than 0.0001, v.s. r = 0.64, p less than 0.0001). Therefore, by measuring the maximal defect diameter, maximal defect area and maximal shunt flow area index, the echocardiography can provide an accurate and non-invasive assessment of isolated VSD in children.