BACKGROUND: A precise definition of ventricular enlargement is important in the diagnosis of hydrocephalus as well as in assessing central atrophy. The Evans index (EI), a linear ratio between the maximal frontal horn width and the cranium diameter, has been extensively used as an indirect marker of ventricular volume (VV). With modern imaging techniques, brain volume can be directly measured. OBJECTIVE: To determine reference values of intracranial volumes in healthy elderly individuals and to correlate volumes with the EI. METHODS: Magnetic resonance imaging (3 T) was performed in 46 healthy white elderly subjects (mean age +/- standard deviation, 71 +/- 6 years) and in 20 patients (74 +/- 7 years) with large ventricles according to visual inspection. VV, relative VV (RVV), and EI were assessed. Ventricular dilation was defined using VV and EI by a value above the 95th percentile range for healthy elderly individuals. RESULTS: In healthy elderly subjects, we found VV = 37 +/- 18 mL, RVV = 2.47 +/- 1.17%, and EI = 0.281 +/- 0.027. Including the patients, there was a strong correlation between EI and VV (R = 0.94) as well as between EI and RVV (R = 0.95). However, because of a wide 95% prediction interval (VV: +/-45 mL; RVV: +/- 2.54%), EI did not give a sufficiently good estimate of VV and RVV. CONCLUSION: VV (or RVV) and the EI reflect different properties. The exclusive use of EI in clinical studies as a marker of enlarged ventricles should be questioned. We suggest that the definition of dilated ventricles in white elderly individuals be defined as VV >77 mL or RVV >4.96 %. Future studies should compare intracranial volumes with clinical characteristics and prognosis.