In this study, a new approach to optimize the cellulose nanocrystals (CNCs) extraction using acidic natural deep eutectic solvents (NADES) was introduced using, for the first time, design of experiment method. Choline chloride:oxalic acid dihydrate with a molar ratio of 1:1 was used to extract CNCs. Then, three most important parameters were varied to design the experiment: (i) cotton fibre concentrations, (ii) temperature and (iii) treatment time. Two outcomes were studied: the CNC yield and the crystallinity. The mathematical model for crystallinity perfectly described the experiments, while the model for CNC yield provided only a tendency. For a reaction time of 6 h at 95°C with a fibre concentration of 2%, the expected optimum CNC yield was approximately 35.5 ± 2.7% with a crystallinity index of 80 ± 1%. The obtained experimental results confirmed the models with 43.6 ± 1.9% and 81 ± 1% for the CNC yield and the crystallinity index, respectively. This study shows that it is possible to predict the CNC yield CNC and their crystallinity thanks to predictive mathematical models, which gives a great advantage to consider in the near future a scale up of the extraction of cellulose nanocrystals using this original family of green solvents.