Abstract: Addressing the issue that batch-to-batch heterogeneity in tobacco flavors affects the sensory quality and combustion performance of tobacco products, and acknowledging the limitations of current quality control methods in proactive production-side regulation, this study aimed to propose a multi-objective optimization method for simultaneously optimizing quality consistency, material cost, and the number of dosing batches in the blending process. This study developed a mathematical blending model based on the Non-dominated Sorting Genetic Algorithm II （NSGA-II）, which incorporated multi-objective functions for critical component deviation, total material cost, and the number of dosing batches, along with constraints such as inventory levels, component thresholds, dosage limits, and batch count restrictions. Using 11 batches of raw materials for a specific tobacco flavor as a case study, the optimization effects under three production scenarios—unrestricted batch count, minimization of batch count as an objective, and constrained batch count—were analyzed and compared. The results indicated that the NSGA-II algorithm generated diverse Pareto optimal sets. Among these, the optimization scheme incorporating explicit constraints on the number of dosing batches demonstrated favorable overall performance in actual production: it reduced the relative deviation of critical components to 1.335% while achieving a sound balance between cost control and operational simplicity. The method proposed in this study helps balance the trade-offs among quality, cost, and production efficiency in tobacco flavor blending, providing scientific decision support and a practical technical pathway to address batch-to-batch quality consistency issues. This conclusion can offer a reference for quality control and efficiency improvement in tobacco flavor production, as well as in other similar industrial manufacturing processes.