Trabecular or cancellous bone, the type of bone found in the vertebrae and near the joints of long bones, consists of a network of plates and struts. Accurate measurement of trabecular thickness is of significant interest, for example, to assess the effectiveness of anabolic (bone forming) agents of patients with osteoporosis. Here, we introduce a new fuzzy distance transform (FDT)-based thickness computation method that obviates binary segmentation and that can effectively deal with images acquired at a voxel size comparable to the typical trabecular bone thickness. The method's robustness is shown on the basis of micro-CT images of human trabecular bone, resampled at progressively coarser resolution and after application of rotation and addition of noise as a means to simulate the in vivo situation. Reproducibility of the method is demonstrated with micro-CT images by comparing histograms of thickness within and between data sets and with micro-MRI volume data sets of human volunteers imaged repeatedly. Finally, with in vivo micro-MR images from a prior study in rabbits subjected to corticosteroid exposure, it is demonstrated that short-term treatment resulting in trabecular thinning can be quantified with the new method.