The utility of M-mode echocardiography in the diagnosis of heart failure (HF) was evaluated in a study of 70 patients with suspected HF (26 men and 44 women) and 63 control persons (26 men and 37 women), all aged 45–74 years. The patients were classified according to the certainty of HF diagnosis using the Boston criteria: 27 patients were defined as ‘unlikely’ to have HF, 19 as having ‘possible’ HF, and 24 as having ‘definite’ HF. In calculations of the sensitivities and specificities for echocardiographic variables in detecting ‘definite’ HF, the 95% confidence limits in the control group were used as cut-off point values. Sensitivities thus attained were 91% for mitral valve E point-septal separation (EPSS), 73% for left ventricular (LV) fractional shortening (FS), and 64% for peak rate of increase of LV diameter (PLR), respectively, and the specificities were 73, 88, and 78%, respectively. When EPSS, FS and PLR were all normal, the likelihood of’definite’ HF was as low as 7%. We conclude that M-mode echocardiography is actually a useful method in the diagnostic evaluation of patients with suspected HF, and it is more reliable in excluding than confirming the presence of HF.