The mechanism by which polyglutamine expansion in Huntington's disease (HD) results in selective neuronal degeneration remains unclear. We previously reported that the immunohistochemical distribution of N-terminal huntingtin in HD does not correspond to the severity of neuropathology, such that significantly greater numbers of huntingtin aggregates are present within the cortex than in the striatum. We now show a dissociation between huntingtin aggregation and the selective pattern of striatal neuron loss observed in HD. Aggregate formation was predominantly observed in spared interneurons, with few or no aggregates found within vulnerable spiny striatal neurons. Multiple perikaryal aggregates were present in almost all cortical NADPH-diaphorase neurons and in approximately 50% of the spared NADPH-diaphorase striatal neurons from early grade HD cases. In severe grade HD patients, aggregates were more prominent as nuclear inclusions in NADPH-diaphorase neurons, with less perikaryal and neuropil aggregation. In contrast, nuclear or perikaryal huntingtin aggregates were present in less than 4% of the vulnerable calbindin striatal neurons in all HD cases. These findings support the hypothesis that polyglutamine aggregation may not be a predictor of cell loss. Rather than a harbinger of neuronal death, mutant huntingtin aggregation may be a cytoprotective mechanism against polyglutamine-induced neurotoxicity.