Mathematical models are very often used in assessing nutritional status and body composition, especially body fat percentage, as they are considered to be an objective indica-tor. However, since every model has its own limitations, models used for assessing nutritional status and/or body composition (body fat percentage) are no exception. Two models for as-sessing nutritional status were used in this thesis (body mass index, waist-hips ratio) and four models for assessing body fat percentage with a malnourished group i.e. anorexia nervosa patients. Malnutrition is the key clinical feature of restrictive anorexia nervosa with associated complications of almost all organs and systems. Patients' conditions were monitored before and after a treatment and the purpose was to establish if there were any statistically significant differences in body fat percentage when different models used. The aim was to establish if there were any significant differences in body fat percentage calculated using different equa-tions. The limit value was set to determine in which situations values would become unac-ceptable for body fat percentage (negative). They varied from model to model. With model 1 and 2 (equations 1 and 2) negative results are a consequence of a logarithm of a skinfold sum (three in model 1 and four skinfolds in model 2) which was less than 6 mm. According to a range method (model 4) the results showed positive values of body fat percentage. The equa-tion uses a logarithm of stomach, hip and neck girth sum. Using t-test it is was determined that there is statistically significant difference in body fat percentage calculated by different equations.