A fuel leaching experiment has been in progress since 1977 to study the dissolution behaviour of used CANDU fuel in aerated aqueous solution. The experiment involves exposure of 50-mm clad segments of an outer element of a Pickering fuel bundle (burnup 610 GJ/kg U; linear and peak power ratings 53 and 58 kW/m, respectively), to deionized distilled water (DDH2O, ∼2 mg/L carbonate) and tapwater (∼50 mg/L carbonate). In 1992, it was observed that the fuel in at least one of the leaching solutions showed some signs of deterioration and, therefore, in 1993, parts of the fuel samples were sacrificed for a detailed analysis of the physical state of the fuel, using SEM and optical microscopy. Leaching results to date show that even after >6900 days only 5 to 7.7% of the total calculated inventory of 137Cs has leached out preferentially and that leach rates suggest a development towards congruent dissolution. Total amounts of 137Cs and 90Sr leached are slightly larger in tapwater than in DDH2O. SEM examinations of leached fuel surface fragments indicate that the fuel surface exposed to DDH2O is covered in a needle-like precipitate. The fuel surface exposed to tapwater shows evidence of leaching but no precipitate, likely because uranium is kept in solution by carbonate. Detailed optical and SEM microscopy examinations on fuel cross sections suggest that grain-boundary dissolution in DDH2O is not prevalent, and in tapwater appears to be limited to the outer %0.5 mm (pellet/cladding) region of the fuel. Grain boundary attack seems to be limited to microcracks at or near the surface of the fuel. It thus appears that grain-boundary attack occurs only near the fuel pellet surface and is prevalent only in the presence of carbonate in solution.