Sixty-six consecutive patients without left ventricular volume overload, significant arrhythmia or significant pericardial effusion were examined by M-mode echocardiography immediately before diagnostic left- and right-heart catheterization. Using various echocardiographic measurements, left ventricular stroke volume (SV) was calculated according to eight different echocardiographic formulas (SVE) that have been proposed previously. At catheterization SV was also determined by thermodilution (SVT) and by single-plane left ventricular cineangiography in the right anterior oblique projection (SVA). When comparing SVE with SVT, the four formulas developed to calculate mitral or aortic flow failed (r = 0.10 to 0.54). As expected, poor correlations (r = 0.22 to 0.47) were also found when formulas used to calculate ventricular volumes from the ventricular diameter or SV from the change in diameter (left ventricular formulas) were used in coronary patients with grossly asymmetrical ventricular contraction patterns. When the use of the left ventricular formulas was confined to patients with symmetrical or almost symmetrical contraction, two formulas yielded favorable correlations of r = 0.84, SEE = 12.7 ml and r = 0.86, SEE = 12.2 ml, respectively. These correlations were comparable to the correlation between our two invasive reference techniques (r = 0.81; SEE = 12.2 ml). The comparison between SVE and SVA confirmed the results of the thermodilution study, though the correlations were generally weaker. We conclude that the formula of Teichholz et al., which was the best of all tested formulas, may be used to obtain a clinically useful estimate of SV in patients in whom symmetrical or almost symmetrical left ventricular contraction can be anticipated.