Huntington's disease (HD) and eight additional inherited neurological disorders are caused by CAG triplet-repeat expansions leading to expanded polyglutamine-sequences in their respective proteins. These triplet-CAG repeat disorders have in common the formation of aberrant intraneuronal proteinaceous inclusions containing the expanded polyglutamine sequences. These aggregates have been postulated to contribute to pathogenesis caused by conformational toxicity, sequestration of other polyglutamine-containing proteins, or by interfering with certain enzymatic activities. Testing these hypotheses has been hampered by the difficulty to isolate these aggregates from brain. Here we report that polyglutamine aggregates can be isolated from the brain of the Tet/HD94 conditional mouse model of HD, by following a method based on high salt buffer homogenization, nonionic detergent extraction, and gradient fractionation. We then verified that the method can be successfully applied to postmortem HD brains. Immunoelectron microscopy, both in human and mouse samples, revealed that the stable component of the inclusions are mutant huntingtin-containing and ubiquitin-containing fibrils. Atomic-force microscopy revealed that these fibrils have a 'beads on a string' morphology. Thus, they resemble the in vitro assembled filaments made of recombinant mutant-huntingtin, as well as the Abeta and alpha-synuclein amyloid protofibrils. Finally, by shutting down transgene expression in the Tet/HD94 conditional mouse model of HD, we were able to demonstrate that these filaments, although stable in vitro, are susceptible to revert in vivo, thus demonstrating that the previously reported reversal of ubiquitin-immunoreactive inclusions does not simply reflect disassembling of the inclusions into their constituent fibrils and suggesting that any associated conformational or protein-sequestration toxicity is also likely to revert.