The aim of this study was to assess microleakage in class V cavities prepared by air abrasion or high-speed dental bur and restored with different glass ionomer cements. Sixty bovine incisors were equally divided into 6 groups: I, II and III (preparation by high-speed) and IV, V and VI (preparation by air abrasion). Groups I and IV were restored with Fuji IX; groups II and V with Ketac Molar; and groups III and VI with Vitremer. After 24 h (37 degrees C), specimens were thermocycled, isolated with nail varnish, immersed in a 0.2% Rhodamine B solution for 24 hours, sectioned longitudinally and analyzed for microleakage using an optical microscope connected to a digital camera and a computer. The images were digitized and a software allowed the quantitative evaluation of microleakage in millimeters. Data were analyzed by Wilcoxon and Kruskal-Wallis tests. It was observed that there were significant differences (p < 0.05) between incisal (enamel) and cervical (dentine/cementum) margins, mainly for Ketac Molar; there was no difference (p > 0.05) between preparation methods, except for group II (high-speed/Ketac Molar) that showed higher infiltration; regarding the materials, Ketac Molar demonstrated the highest microleakage values (p < 0.05), and only Vitremer sealed completely both margins of restorations. It was concluded that air abrasion preparation did not influence microleakage in class V restorations with the employed glass ionomer cements.