The polyglutamine disorders consist of a group of nine neurodegenerative diseases with overlapping phenotypes, but which affect distinct neuronal subsets, causing neuronal dysfunction and death. In the majority of these, the causative proteins share no homology to other known proteins, or to each other apart from the polyglutamine tract. The polyglutamine tracts themselves are toxic over a disease-specific threshold, and this common feature has suggested a common pathogenesis. The pathogenic mechanism(s) of this group of diseases is hotly debated, with proteolytic cleavage and nuclear accumulation both popular hypotheses. Such cleavage is thought to release toxic fragments containing an expanded polyglutamine tract, and may itself facilitate entry of cytoplasmic polyglutamine proteins to the nucleus. Numerous downstream effects including accumulation and apoptotic activation, misfolding, aggregation, and sequestration of other proteins including transcription factors and chaperones may then be initiated. It is uncertain whether all of the polyglutamine proteins undergo cleavage in vivo. Even in those in which proteolysis has been demonstrated, it remains unclear to what extent this also occurs in the wild-type proteins, or whether it is dependent on, or increased by, the expanded polyglutamine tract. Similarly, in at least one of these disorders (spinocerebellar ataxia type 6), nuclear localisation has not been demonstrated. The contradictory evidence for the production and role of proteolytic fragments and for nuclear localisation in toxicity, reviewed in this article, suggests that neither may be uniformly necessary steps in the pathogenesis of this group of diseases, and that, for all their apparent similarities, the exact pathogenic mechanisms may not be identical in each.