In active metal brazing, the reactive element in the brazing alloy enhances the wettability of the brazing alloy on ceramics and causes strong bonding by the redox reaction [1, 2]. The joint quality is also a function of the interfacial microchemistry and microstructures, such as the species and the morphology of the reaction product formed at the ceramic=brazing alloy interface [3, 4]. Therefore, it is very important to characterize and understand the interfacial reaction and its product at the metal= ceramic interface. However, in active metal brazing, there are many possible reactions at the interface between the reactive element and the ceramic or brazing alloy together with the bonded metal. Because of these complex reactions, the interfacial reaction is difficult to control. There are many reports on metal-ceramic joining with Ti containing brazing alloy, but few attempts have been made in brazing of alumina to metal with Zr contained AgCu based filler metal nor investigations of the interfacial reaction. In this study, Zr was selected as an active element for the brazing alloy since it showed stronger work of adhesion (Wad) on Al2O3 ceramics and more negative G of oxide formation [5] than Ti. The interfacial reaction and the effect of the morphological change of the brazing on the joint strength are discussed here. Discs of 99.9% AE-Al2O3 (diameter 14.5 mm and thickness 5 mm) were brazed to Ni-Cr steel and copper (diameter 12 mm and thickness 8 mm) using three kinds of brazing alloys, Ag-28 wt % Cu 5 wt % Zr (BZR), BZR 5 wt % Sn (BZS) and BZR 5 wt % Al (BZA). The brazing alloys were made by vacuum induction melting. Prior to brazing, all materials used were prepared in the same way as described elsewhere [6]. An elliptical radiant furnace equipped with a quartz tube was employed in the brazing experiment and it was evacuated to less than 10y5 torr (1 torr 133.322 Pa). Brazing was carried out in the temperature range of 750– 950 8C and the holding time was 30 min. A shear test was performed to evaluate the joint strength, and the microstructure and the interfacial reaction product were analysed using electron probe X-ray micro-analysis, EPMA, (JEOL JXA 840A) equipped with an energy dispersive X-ray spectroscopy EDS, (Link analytical LZ5) and glancing X-ray diffraction, XRD (Regaku Rotaflex RTP 300RE) and X-ray photoelectron spectroscopy, XPS (SSI 2830-S). Fig. 1 shows the SEM microstructure of (a) the interfacial structure and the concentration profile and (b) the reaction layer for Al2O3=Ni-Cr steel joint brazed at 950 8C for 30 min with Ag 28 wt % 5 wt % Zr brazing alloy. As depicted in Fig. 1b, the reaction layer consisted of Zr and small amounts of Al and Ag. The low Al content in the reaction product might be a result of the higher diffusivity of Al compared to that of oxygen in the molten brazing alloy and an unfavourable reaction to produce Zr-Al compound as compared with Zr-oxide. XRD analysis indicated that the reaction product was monoclinic ZrO2 (a 5:1463, b 5:2135, c 53110 A) even though EDS results showed the existence of Ag in the reaction product. However, Al2O3 is a more stable oxide thermodynamically in comparison with ZrO2 hence it is not feasible to form ZrO2 from zirconium and Al2O3. Simple equilibrium thermodynamics can not sufficiently characterize the interfacial reaction phenomena in the present brazing system. The Ag