Patients with polyglutamine expansion diseases, like Huntington's disease or several spinocerebellar ataxias, first present with neurological symptoms that can occur in the absence of neurodegeneration. Behavioral symptoms thus appear to be caused by neuronal dysfunction, rather than cell death. Pathogenesis in polyglutamine expansion diseases is largely viewed as a cell-autonomous process in neurons. It is likely, however, that this process is influenced by changes in glial physiology and, at least in the case of DRPLA glial inclusions and glial cell death, seems to be an important part in the pathogenesis. To investigate these aspects in a Drosophila model system, we expressed polyglutamine proteins in the adult nervous system. Glial-specific expression of a polyglutamine (Q)-expanded (n=78) and also a nonexpanded (n=27) truncated version of human ataxin-3 led to the formation of protein aggregates and glial cell death. Behavioral changes were observed prior to cell death. This reveals that glia is susceptible to the toxic action of polyglutamine proteins. Neuronal expression of the same constructs resulted in behavioral changes similar to those resulting from glial expression but did not cause neurodegeneration. Behavioral deficits were selective and affected two analyzed fly behaviors differently. Both glial and neuronal aggregates of Q78 and Q27 appeared early in pathogenesis and, at the electron microscopic resolution, had a fibrillary substructure. This shows that a nonexpanded stretch can cause similar histological and behavioral symptoms as the expanded stretch, however, with a significant delay.