A unifying feature of the CAG expansion diseases is the formation of intracellular aggregates composed of the mutant polyglutamine-expanded protein. Despite the presence of aggregates in affected patients, the precise relationship between aggregates and disease pathogenesis is unresolved. Results from in vivo and in vitro studies of mutant huntingtin have lead to the hypothesis that nuclear localization of aggregates is critical for the pathology of Huntington's disease (HD). We tested this hypothesis using a 293T cell culture model system that compared the frequency and toxicity of cytoplasmic and nuclear huntingtin aggregates. We first assessed the mode of nuclear transport of N-terminal fragments of huntingtin, and show that the predicted endogenous NLS is not functional, providing data in support of passive nuclear transport. This result suggests that proteolysis is a necessary step for nuclear entry of huntingtin. Additionally, insertion of nuclear import or export sequences into huntingtin fragments containing 548 or 151 amino acids was used to reverse the normal localization of these proteins. Changing the subcellular localization of the fragments did not influence their total aggregate frequency. There were also no significant differences in toxicity associated with the presence of nuclear compared with cytoplasmic aggregates. The findings of nuclear and cytoplasmic aggregates in affected brains, together with these in vitro data, support the nucleus and cytosol as subcellular sites for pathogenesis in HD.