The objective of this in vitro study was to compare resistance to enamel demineralization after banding with 3 orthodontic cements. The 3 orthodontic cement groups and the nonbanded control group were evaluated for cariopreventive potential. One hundred twenty extracted human molars were selected for banding, embedded in resin blocks, and randomly assigned to the following 4 groups: zinc phosphate cement, zinc polycarboxylate cement, resin-modified glass ionomer (RMGI), and nonbanded control. Orthodontic bands were placed and cemented; specimens were stored in artificial saliva at 37 degrees C for 30 days, thermocycled for 24 hours, and then debanded with a customized band-removal device attached to a universal testing machine. The cement was hand removed, and 2 coats of nail varnish were applied to the teeth, leaving a 2 x 2-mm window exposed on the buccal surface of the mesiobuccal cusp. The teeth were stored in lactic acid-gelatin solution for 4 weeks at 37 degrees C to develop simulated white spot decalcification in the window. The teeth were subjected to dye (10% methylene blue) for 24 hours and then sectioned through the window. The depth of dye penetration was determined and used as a measure of the cariopreventive effect of the banding cement. Data were analyzed with a 1-way ANOVA and a Tukey test for the multiple comparisons. Dye penetration occurred as follows: the zinc phosphate and control groups were about the same; there was less dye penetration with the zinc polycarboxylate, and the least with the resin-modified glass ionomer. The 2 fluoride-releasing cements (zinc polycarboxylate and RMGI) demonstrated less demineralization than the zinc phosphate in vitro and might provide greater protection from demineralization around a band periphery in clinical settings.