Modeling of soil shrink-swell requires accurate measurement of volume changes in soil samples during drying-wetting cycles. Soil core samples are needed if anisotropy in shrinkage behavior has to be evaluated. The widely used caliper method allows soil core volume measurements but shows difficulties in providing accurate evaluation of horizontal deformation and risk of sample disturbance, particularly in wet conditions. Most recently proposed techniques (e.g.: laser, paraffin) either exhibit limitations in terms of complexity and equipment costs or are long and laborious. In this work an alternative method for accurately determining soil core volume, which is faster and relatively more simple and cheaper, is proposed. This method is based on simple 2D image analysis of two radiographs of the sample, whose results allow applying Pappus’ 2nd centroid theorem to calculate the volume of the soil core. A validation test performed by comparison with the caliper method has shown an error in accuracy lower than 1% for the proposed method versus 5% for the caliper method. The radiographic method has been more precise on a reference sample (standard deviations about 1.0% versus 1.4% of the average volume value) and also on four repacked cores of natural soils with different shrink-swell potentials (standard deviations ranged from 0.39% to 1.15% and from 0.34% to 3.00% of the volume for the radiography and caliper methods, respectively). Application of the method to the same soil samples during a drying process has highlighted its greater accuracy and suitability in assessing anisotropy of shrinkage at any soil water content.