Most neurodegenerative diseases are linked to aberrant accumulation of aggregation-prone proteins. Among them, Huntington's disease (HD) is caused by an expanded polyglutamine repeat stretch in the N terminus of the mutant huntingtin protein (mHTT), which gets cleaved and aggregates in the brain. Recently established human induced pluripotent stem cell-derived HD neurons exhibit some disease-relevant phenotypes and provide tools for HD research. However, they have limitations such as genetic heterogeneity and an absence of mHTT aggregates and lack a robust neurodegeneration phenotype. In addition, the relationship between the phenotype and mHTT levels has not been elucidated. Herein, we present a human embryonic stem cell (hESC)-derived HD neuronal model expressing HTTexon1 fragments, which addresses the deficiencies enumerated above. The wild-type and HD lines are derived from an isogenic background and exhibit insoluble mHTT aggregates and neurodegeneration. We also demonstrate a quantitative relationship between neurodegeneration and soluble monomeric (but not oligomeric or aggregated) mHTT levels. Reduction of ~10% of mHTT is sufficient to prevent toxicity, whereas ~90% reduction of wild-type HTT is safe and well-tolerated in these cells. A known HD toxicity modifier (Rhes) showed expected rescue of neurodegeneration. Therefore, the hESC-derived neuronal models complement existing induced pluripotent stem cell-derived neuronal models and provide valuable tools for HD research.