In order to optimize the quality of vermicelli of purple potato starch, the formula of five kinds of starch from purple potato, sweet potato, tapioca, potato and purple potato residue powder, were investigated with the D-optimal mixture design. The principal component analysis and dimensionless normalization method were used to comprehensively assess the quality of vermicelli from purple potato starch. During the mixture design experiment, the linear regression models were built based on the response values of the three synthesis scores of the texture characteristics (T₁), cooking characteristics (T₂) and sensory characteristics (T₃). The results indicated that the effects of the five components and their interactions on the texture characteristics (T₁) and sensory characteristics (T₃) were signicant（R²=0.921 1，P＜0.001；R²=0.883 8，P=0.005 3）. According to the relative results of indicators correlation analysis and the model fitting, T₁ and T₃ values were selected as the optimal indices of the mixture design. The final optimized formula was found to contain the different kinds of starch with 35.0% from purple potato, 15.0% from sweet potato, 18.5% from tapioca, 9.5% from potato, and 22.0% from purple potato residue powder. This new vermicelli of purple sweet potato tasted smooth with bright color. The research provided a technology support for the processing new type of purple potato vermicelli.