Glass transition temperatures (Tg) of amorphous sugar samples, prepared from 0% to 14% (wb) moisture content, were determined during both cooling and heating using differential scanning calorimetry (DSC). The DSC protocol used involved heating, cooling, and reheating the sugar samples, where Tg was obtained during both cooling and reheating steps. The Gordon-Taylor equation modeled the water plasticization effect, where the Tg of anhydrous solids and constant (k) values were determined simultaneously. No significant difference was found between the cooling and heating glass transition temperature values, with both values significantly influenced by sugar composition (p < 0.05). No significant effect of sugar composition was observed for the k values, resulting in discrepancies displayed by the Gordon-Taylor model for solid mass fractions below 0.84. Thus, k values should be estimated from measured Tg data at low solid mass fractions rather than extrapolated from high solid mass fractions, for prediction of Tg curve at high moisture contents. (C) 2014 Elsevier Ltd. All rights reserved.