BackgroundUnder-nutrition in neonates is closely linked to low body fat percentage. Undernourished neonates are exposed to immediate mortality as well as unwanted health impacts in their later life including obesity and hypertension. One potential low cost approach for obtaining direct measurements of body fat is near-infrared (NIR) interactance. The aims of this study were to model the effect of varying volume fractions of melanin and water in skin over NIR spectra, and to define sensitivity of NIR reflection on changes of thickness of subcutaneous fat. GAMOS simulations were used to develop two single fat layer models and four complete skin models over a range of skin colour (only for four skin models) and hydration within a spectrum of 800–1100 nm. The thickness of the subcutaneous fat was set from 1 to 15 mm in 1 mm intervals in each model.