Normal-incidence ultrasonic. scans have been conducted on two- and three-layer adhesive-joint specimens exposed to water at 50degreesC for periods of up to 18 months. The joints consisted of aluminium-alloy adherends which were subjected to one of four different surface pretreatments prior to being coated (for the two-layer specimens) or bonded (for the three-layer specimens) with an epoxy polymer. The four commonly used pretreatments which were investigated were a grit-blast, a chromic-acid etch (CAE), a chromic-acid anodise (CAA) and a phosphoric-acid anodise (PAA). Techniques have been developed to measure the fracture toughness, G(c), of the specimens before and after water exposure such that fracture toughness maps could be ascertained, where the measured values of G(c) may be assigned to spatially discrete positions within the specimens. The relative performance of the different pretreatments used in the present work followed that expected from the literature: the chromic-acid anodised (CAA) surface pretreatment giving the most durable two- and three-layer specimens and the grit-blasting pretreatment giving the least durable. In the two-layer specimens, the ultrasonic inspections detected two main types of defects: corrosion-driven edge-disbonds and micro-defects. The edge-disbonding mechanism usually started at an unsealed flush edge, and was initiated by a region of corrosion which developed on the edge of the specimen and which undercut the epoxy layer. Edge-disbonds were easily and accurately detected ultrasonically. Micro-defects were detected in regions remote from the edges and these small-scale, isolated defects took several forms. In the case of the three-layer joint specimens, only edge-disbonds could be detected ultrasonically. Nevertheless, in some of the three-layer joints which were attacked and weakened by ingressing water, the failure surfaces suggested that micro-defects were present. For both the two- and the three-layer specimens, the results from the ultrasonic scans have been correlated with the values of the fracture toughness of the specimens, before and after water exposure. Whilst the ultrasonic scans detected the presence of micro-defects in the two-layer specimens, which appeared to correlate with the extent of interphase toughness loss upon water exposure. the scans clearly failed to detect any changes in the interphase regions which would indicate the general loss of interphase toughness seen with the three-layer grit-blast and PAA specimens.