A fracture mechanics approach has been successfully used to examine the cyclic fatigue behaviour of adhesively-bonded joints, which consisted of aluminium-alloy or electro-galvanised (EG) steel substrates bonded using toughened-epoxy structural paste-adhesives. The adhesive systems are typical of those being considered for use, or in use, for bonding load-bearing components in the automobile industry. The results were plotted in the form of the rate of crack growth per cycle, da/dN, versus the maximum strain-energy release rate, G(max) applied in the fatigue cycle, using logarithmic axes. Of particular interest was the presence of a threshold value of the strain-energy release rate, G(th), applied in the fatigue cycle, below which fatigue crack growth was not observed to occur. The cyclic fatigue tests conducted in a relatively dry environment of 23 degrees C, and 55%; RH were shown to cause crack propagation at far lower values of G(max) compared with the value of the adhesive fracture energies, G(c), which were determined from monotonically-loaded fracture tests. Cyclic fatigue tests were also conducted in a "wet" environment, namely immersion in distilled water at 28 degrees C. The "wet" fatigue tests clearly revealed the further significant effect an aggressive, hostile environment may have upon the mechanical performance of adhesive joints, and highlighted the important influence that the surface pretreatment, used for the substrates prior to bonding, has upon joint durability. The development and standardisation of "wet" fatigue tests may provide the basis for a very effective accelerated-ageing test.