Memory plays a major role in the formation of food expectations. How accessible and how accurate is incidentally acquired and stored product information? In the present experiment the memory for variations in texture (and flavour) was tested with a new and ecologically valid method. Subjects (N = 69:35 women, 34 men, age 19-59 yrs) came to the institute before having had breakfast and received a breakfast of several items (including breakfast drinks, biscuits and pâté). Subsequently, during the day, they answered questions about their hunger feelings every hour and returned for a taste experiment at the end of the day. When unexpectedly confronted with a series of multiple samples of four texture variations of each of the breakfast items mentioned above, they had to indicate which of these variations they had eaten at breakfast. Signal detection measures showed that most subjects recognised the eaten versions well for all three of the food items. Women remembered better than men. Both men and women used an equally conservative criterion. © 2002 Elsevier Science Ltd. All rights reserved.