Thermograms of wheat, corn and rice flour samples were recorded using pressure variable DSC. Pressure was maintained at 20 atm during measurements, and thermograms were normalized to yield specific heat curves as a function of temperature, moisture content and protein content. Specific heat data were fitted to a general empirical equation, using an algorithm for multiple regression with subsequent statistical analysis. A quantitative empirical equation incorporating temperature, moisture content, and protein content dependent terms was developed to predict specific heat of cereal flours and starches between 20 and 110 degrees C and the moisture content between 0% and 70% dry basis. Proposed equation predicts the specific heat of flours in the database with an average absolute error of 2.3% and the specific heat of cereal flours and starches reported in the literature with an average absolute error of 4.3%. (C) 2007 Elsevier Ltd. All rights reserved.