The quality and marketability of many fried food products is a function of the internal distribution of oil and water. French fries, for example, are well received when they are both crisp and moist, but not when they are soggy. A rapid technique for the measurement of local oil and water content is therefore highly desirable to guide the development and processing of fried food materials. Time-domain nuclear magnetic resonance (TD-NMR) has proven to be a quick, reproducible, accurate, and non-invasive technique, which is particularly well suited to measurements of oil and moisture content. Nuclear spin-spin relaxation times (T-2) are an excellent probe of molecular mobility, which in turn can be directly correlated to Moisture content. In this work, a precursor to spatially resolved measurements, we measure T-2 in a model fried food crust sample as a function of moisture level. We observe multi-exponential behavior, which we assign to specific liquid constituents. We find a direct correlation between the component weightings and the known oil and water contents. By systematically varying the moisture level of the sample, we explore the nature of water uptake in these materials. Absorption is modeled with a modified BET sorption isotherm. These results provide a basis upon which magnetic resonance imaging (MRI) techniques can be applied to investigate cooked starch food crusts. (c) 2006 Elsevier Ltd. All rights reserved.