We constructed a liquid phantom, which mimics the neonatal head for testing near infrared spectrophotometry instruments. It consists of a spherical, 3.5 mm thick layer of silicone rubber simulating skin and bone and acts as container for a liquid solution with IntralipidTM, 60 micrometers ol/l haemoglobin and yeast. The IntralipidTM concentration was varied to test the influence of scattering on haemoglobin concentrations and tissue oxygenation determined by the Critikon 2020. The solution was oxygenated using pure oxygen and then deoxygenated by the yeast. For the instruments algorithm, we found with increasing scattering (0.5%, 1%, 1.5% and 2% IntralipidTM concentration) an increasing offset added to the oxy- (56.7, 90.8, 112.5, 145.2 micrometers ol/l respectively) and deoxyhaemoglobin (25.4, 44.3, 58.5, 65.9 micrometers ol/l) concentration causing a decreasing range (41.3, 31.3, 25.0, 22.2%) of the tissue oxygen saturation reading. However, concentration changes were quantified correctly independently of the scattering level. For an other algorithm based on the analytical solution the offsets were smaller: oxyhaemoglobin 12.2, 34.0, 53.2, 88.8 micrometers ol/l and deoxyhaemoglobin 1.6, 11.2, 22.2, 28.1 micrometers ol/l. The range of the tissue oxygen saturation reading was higher: 71.3, 55.5, 45.7, 39.4%. However, concentration changes were not quantified correctly and depended on scattering. This study demonstrates the need to develop algorithms, which take into consideration the anatomical structures.