Since the discovery of acrylamide in heat-treated carbohydrate-rich foods, many more heat-generated food contaminants have been identified in a variety of foods and models systems. A database of these contaminants, generated as a result of either lipid oxidation or the Maillard reaction, has recently been compiled under the HEATOX project. A large majority of the compounds has not been tested for potential adverse effects on human health, which makes it difficult to carry out adequate assessment of risks to an average consumer. This study used two in silico toxicity Expert Systems (Topkat and Derek for Windows), as a preliminary screening tool to identify potential toxicants among the heat-generated contaminants in foods or model systems. The methodology enabled prioritisation of the compounds on the basis of predicted toxicities, and identification of potential toxicants for targeted testing by standard laboratory procedures. A comparison between the predicted toxicities of selected compounds and the available experimental data indicated that the methodology can be reliably used for assessing toxicity of untested food contaminants.