SummaryNormal-and oblique-incidence ultrasonic scans have been conducted on two-and three-layer adhesive-joint specimens exposed to water at 50 o C for periods of up to 18 months. The joints consisted of aluminium alloy adherend(s) which was subjected to one of four different surface pretreatments prior to being coated (for the two-layer specimens) of bonded (for the three-layer specimens) with an epoxy polymer. Overall, the oblique-incidence technique detected the same types of defect as the normal incidence scans, except when the resolution was too poor to detect the smallest defects. No defect was detected using oblique-incidence scans that could not be detected using normal-incidence scans. In the two-layer specimens, two main types of defects were detected: edge disbonds and micro-defects. Micro-defects were detected in regions remote from the edges and these small-scale, isolated defects appeared took several forms. In the case of the two-layer specimens, the detection of micro-defects was clearly a strong indicator that the interphase of the specimen had been attacked and weakened by the ingressing water molecules and, indeed, the number of such defects correlated to the loss of interfacial toughness. In the case of the three-layer specimens, the extent of edge disbonding was much lower than compared to the corresponding twolayer specimens. Also, no micro-defects were detected ultrasonically in any of the three-layer specimens. However, examination of some of the failure surfaces from the three-layer specimens suggested that some micro-defects may have been present but that they were too small to be detected at the spatial resolution of the ultrasonic scans. Therefore, for the three-layer joint specimens, there were no indicators from the ultrasonic scans which could reveal whether the joint had suffered attack and weakening of the interphase regions by the ingressing moisture. Finally, for all the tests conducted it was very noteworthy that there was no evidence of a gradual change in either the normal-or oblique-incidence reflection coefficient moving into an apparently wellbonded region from either the edge disbonds or the micro-defects, when present.