The investigation of left ventricular contractile reserve usually requires the determination of left ventricular volume, but its measurement with radionuclide angiography is difficult. The aim of this study was to determine left ventricular volume directly during exercise by the simultaneous measurement of peak exercise left ventricular ejection fraction (LVEF) and oxygen consumption (V over dot O-2max) and to compare the results with another geometric method. In the absence of lung disease, the systemic arteriovenous oxygen difference (DAV(max)) during maximal exercise converges to 0.13-0.14 ml O-2 per ml blood. The measurement of V over dot O-2max allows maximal cardiac output (COmax) to be calculated as V over dot O-2max = COmax. DAV(max). By simultaneously determining LVEF(ex), exercise end-diastolic volume (EDV(ex)) can then be expressed as a linear function of V over dot O-2max, maximal heart rate (HR(max)), DAV(max) and LVEF(ex). Then, the relationship between end-diastolic counts and true volume can be derived at rest. The two methods were closely correlated (r = 0.91, P < 0.001), despite the geometric method being less accurate when applied to low counting statistic acquisitions. We conclude that rest and exercise left ventricular volume can be determined non-invasively by the simultaneous measurement of V over dot O-2max and LVEF(ex). Furthermore, this method provides additional prognostic information which is clinically relevant in the staging of patients with heart failure.