Assessment of right ventricular (RV) function is clinically relevant in the follow-up of various forms of congenital heart disease. Agreement on the value of different echocardiographic approaches for this purpose is lacking. Magnetic resonance imaging (MRI) provides dimensionally accurate RV volumes and ejection fraction. Transthoracic 2-dimensional echocardiography from 3 different views and gradient-echo tomographic MRI were performed in 16 children with congenital heart disease and 17 age-matched healthy children. RV volumes and ejection fraction were calculated with 5 mono- and biplane area-length and multiple-slice echocardiographic methods. Adequate MRI and echocardiographic apical 4-chamber images could be obtained in all 33 children. The best correlation between MRI and echocardiographic volumes was with the biplane pyramidal approximation method. End-diastolic volume by MRI was 92 +/- 27 ml: systematic difference with echocardiography was +14 +/- 16 ml (r = 0.86). End-systolic volume by MRI was 33 +/- 13 ml: systematic difference with echocardiography was -4 +/- 7 ml (r = 0.82). Ejection fraction by MRI was 65 +/- 8%: systematic difference with echocardiography was +5 +/- 7% (r = 0.72), using monoplane ellipsoid approximation. For all echocardiographic methods, significant effects of RV geometry were noted. Echocardiographic mono- and biplane area-length and multiple-slice calculations demonstrated moderate correlation and significant systematic errors compared with MRI-derived RV volumes. Echocardiographic results were influenced by RV geometry. The relatively simple monoplane area-length method provides ejection fraction results acceptable for clinical practice; results are not improved by more complex biplane and/or multislice methods.