Some fuel assemblies containing pins manufactured by British Nuclear Fuels Limited failed during irradiation in the Dodewaard Boiling Water Reactor. At discharge, the assemblies had accumulated a mean burnup of 14 870 MWd/Te(U) [14.87 MWd/kg(U)]. A selection of failed and unfailed pins from two of these assemblies was examined by the Central Electricity Generating Board to locate the primary failure sites and to identify the failure mechanism. Eddy-current signals not attributable to any visible feature were observed near the bottom grid site of the seven pins identified as failures. Metallographic examination of this region of four of these pins revealed a primary failure in the form of a penetrating crack in the cladding. It was inferred that the eddy-current signals from the remaining three failed pins originated at similar sites. The failure characteristics were identical to those known to have been caused by power ramps. Furthermore, increases in turbine off-gas and coolant iodine activities were coincident with large power increases at the failure location caused by movement of control blades. It was therefore deduced that the failure of these pins was a consequence of power ramping. A nonpenetrating crack that was not detected by eddy-current testing was found in the unfailed pin that experienced the greatest increase in power. The characteristics of this crack were the same as those found in failed pins. This is regarded as further evidence that the primary source of failure had been located in the failed pins. Several other instances of clad penetration and an end plug failure were observed that were caused by hydriding of the cladding following coolant ingress at the site of the primary failure. Although severe oxidation and associated metal loss were observed at grid positions on most pins, no evidence of clad penetration by this mechanism was found.