In clinical practice, geriatric nutritional assessment usually includes nutritional screening, a simple anthropometric assessment, measurement of various biochemical parameters, such as serum albumin, and sometimes (not always) body composition analysis (BCA). However, there is a high prevalence of undiagnosed malnutrition in patients with dementia. Several factors contribute to this situation; probably, the most notable is the methodology used to assess body composition (BC). In this regard, for BCA, techniques are needed that are noninvasive, affordable, safe, simple and that require the minimum possible collaboration by the elderly patient. Consequently, body mass index (BMI) and waist circumference (WC) are widely used as indicators of overall and central adiposity, respectively; however, there is no consensus on the cutoffs for the elderly, and changes in BC (especially muscle-mass depletion) are masked by normal values of BMI and WC. Bioimpedance analysis is a simple, cost-effective and precise method for BCA, provided that cross-validated equations are used. Its main disadvantage is that it is highly sensitive to changes in body water (overhydration or dehydration), leading to substantial errors in BC estimates. However, using Bioelectrical Impedance Vector Analysis errors are minimized, as there is no need for the subject to be normally hydrated and it does not require the use of predictive models.