Despite the fact that the gene responsible for Huntington's disease (HD) is known, we still do not understand the underlying mechanisms leading to neurodegeneration and death. Identifying and understanding the mechanisms controlling mutant huntingtin (mHtt) aggregation and inclusion formation using different cellular and animal models is crucial to elucidate the molecular mechanisms underpinning the disease and to develop effective treatments to prevent or slow the progression of HD. At the mechanistic level, our work shows that mHtt aggregation and inclusion formation in the cytosol and nucleus occur via different mechanisms and lead to the formation of inclusions with distinct biochemical and ultrastructural properties. We show that mHtt cytoplasmic inclusion formation occurs via two phases: 1) a first phase involves the rapid formation of a dense fibrillar core and is driven predominantly by intermolecular interactions involving the polyQ domain via phase separation-like mechanisms; 2) a second phase is associated with the recruitment of soluble mHtt, fibril growth, and the active recruitment and sequestration of lipids, proteins, and membranous organelles. Our work points to this second phase as an important contributor to mHtt toxicity and suggests that targeting inclusion growth and maturation represent a promising therapeutic strategy. These observations suggest that the two types of inclusions may exert their toxicity via different mechanisms and may require different strategies to interfere with their formation, maturation and toxicity. Using primary neurons, we demonstrated that neuronal intranuclear inclusions evolve over time from small aggregates to large granulo-filamentous inclusions associated with cellular toxicity. In addition, our work underscores the failure of the cellular degradation machineries to clear mHtt inclusions, as well as the sequestration of other proteins, lipids and organelles. Finally, we would like to emphasize that our comparative analysis of tag-free and GFP-tagged mutant mHtt aggregation and inclusions formation will inform future efforts to develop models that reproduce HD pathology more faithfully and underscore the need for developing label-free techniques to investigate disease-relevant mechanisms that underpin inclusion formation.