INTRODUCTION: To reduce microleakage in high-copper amalgam restorations, bonding of amalgam was introduced. AIM: This study compared the microleakage of admixed and spherical amalgams when bonded with different bonding intermediates under thermo- and non-thermocycling conditions. METHOD AND MATERIALS: Class II butt-joint cavities were prepared in 200 extracted human molar teeth, and randomly divided into 5 groups. Calibra, Duo Cement Plus, RelyX ARC and Amalgambond Plus were applied to 4 of these groups. The fifth group was left untreated. The groups were further divided and restored with either Dispersalloy or Oralloy Magicap S. Ten specimens of each group were thermocycled between 5 degrees C and 55 degrees C, placed in basic fuchsin for 8 hours, sectioned and evaluated for dye penetration under 40X magnification. The mean microleakage scores were analysed using the chi-squared test at a confidence level of 95%. RESULTS: Microleakage of the non-bonded amalgams was significantly higher (p < 0.0001) than for the bonded amalgams (thermocycled and non-thermocycled). The microleakage of the different intermediates bonded to Dispersalloy (thermocycled and non-thermocycled) was not significantly different (p > 0.05). The microleakage of the different intermediates was not significant different except for Duo Cement compared to Calibra (p < 0.0001), RelyX (p < 0.0001) and Amalgambond (p = 0.0433) and Amalgambond compared to Calibra (p = 0.0433) and RelyX (p = 0.0433). The microleakage of the bonded amalgams was not significantly increased by thermocycling (p > 0.05). The microleakage of the two amalgams when bonded with the same resin cements (thermocycled and non-thermocycled) was not significantly different except for Duo Cement (thermocycled) (p = 0.0051) and RelyX (non-thermocycled) (p = 0.0356). CONCLUSIONS: Bonding amalgam restorations to tooth structure in butt-joint cavities will reduce microleakage of both admixed and spherical amalgam restorations. Thermal stress does not affect the bond adversely.