Recently, we introduced an intra-oral dentin demineralization model to evaluate the efficacy of a fluoridated toothpaste at plaque-retention sites with limited access to salivary components and fluoride. Usage of the fluoridated toothpaste reduced mineral loss from grooves in sound dentin by 15% compared with a non-fluoridated toothpaste (Lagerweij et al., 1997). In the current study, preformed dentinal lesions were subjected to similar toothpaste treatments. To improve the power of the model in discriminating between toothpaste groups, we changed the design from monadic to crossover to reduce the influence of variation between the participants. Two sets of specimens were placed in the partial prostheses of 29 participants who used a toothpaste containing either 1000 ppm fluoride or no fluoride during two consecutive six-week periods. Changes in mineral content were analyzed by transverse microradiography. The average mineral loss after the nonfluoride period was 1785 vol% × μm (SD 2399), and 1335 vol% × μm (SD 2039) after the fluoride period, a 25% inhibition in demineralization. Analysis of variance showed that the demineralization was significantly influenced not only by the type of treatment (p < 0.001 level), but also by the participant (p < 0.001), the width of the groove (p < 0.001), and the depth into the groove where changes in mineral content were measured (p < 0.017). Twenty-one percent of all the lesions showed remineralization. No significant difference in the extent or the occurrence of remineralization was found between the treatments.