BACKGROUND: Ejection fraction (EF) is commonly applied as a clinically relevant metric to assess ventricular function. The numerical value of EF depends on the interplay between end-systolic volume (ESV) and end-diastolic volume (EDV). Remarkably, the relative impact of the two constitutive components on EF received little attention. METHODS: Three patient groups not using beta-blockers were analyzed for a robust investigation into the relative contribution of ESV and EDV when assessing EF: cardiac patients (N=155) with left ventricular (LV) data obtained by biplane ventriculography, near-normals (N=276) by gated SPECT investigation, and an MRI-based post Fallot repair study including right ventricular (RV) data (N=124), besides LV. We compared various routes to evaluate EF via linear and several types of nonlinear regression with ESV as independent variable. Advanced statistics was applied to evaluate sex-specific differences. RESULTS: In all cases ESV emerges as the dominant component of EF, with less (P<0.0001) impact of EDV. The relationship for EF versus ESV is nonlinear (P<0.0001), and similar for both sexes. A linear approach may be inadequate and generate erroneous statistical outcomes when comparing subgroups of patients. CONCLUSIONS: Values for EF primarily depend on ESV, both for LV and RV. This relationship is essentially nonlinear, and similar for both sexes. A logarithmic approximation is convenient and often acceptable. However, application of linear regression for EF vs ESV may lead to incorrect conclusions, particularly when comparing males and females.