The left ventricular end-systolic pressure-volume relation is a relatively load-independent measure of left ventricular contractile function. Linearity of the relation derived from full left ventricular pressure-volume loops has not previously been demonstrated for patients with severe heart failure. Therefore, nine patients with markedly depressed left ventricular systolic function (ejection fraction 0.14 +/- 0.08) were studied with micromanometer left ventricular pressure measurement and simultaneous radionuclide ventriculography. Afterload was reduced with graded infusions of nitroprusside, allowing construction of pressure-volume loops under four afterload conditions in four patients and three afterload conditions in the other five patients. The end-systolic pressure-volume relation derived from the pressure-volume loops was found to be linear for the range of pressures and volumes examined, with correlation coefficients in individual patients ranging from 0.936 to 0.999 (mean 0.981). The mean slope of the relation (or end-systolic elastance) was 0.71 mm Hg/ml (range 0.42 to 1.52), and the extrapolated volume intercept at zero pressure was positive in all patients. An exponential relation between end-systolic elastance and ejection fraction was demonstrated for this group of patients. Approximations of end-systolic elastance obtained from measurements other than the full pressure-volume loops correlated variably with "true" elastance obtained from the pressure-volume loops. The relation between stroke work and end-diastolic volume was nonlinear in most patients. Thus, the end-systolic pressure-volume relation is linear in the "physiologic" range in patients with severe heart failure. This finding should permit construction of the relation from two loading conditions in clinical studies, facilitating its use as an index of contractile function in patients with heart failure.