<p>Cost sensitivity is a critical issue in sequential multi-class three-way decision-making, aiming to make reasonable decisions at an appropriate costs. However, existing studies lack a comprehensive consideration of both decision result cost and decision process cost when evaluating attribute importance and making decisions. In this paper, we propose a novel sequential multi-class three-way decision model by optimizing the evaluation of attribute importance and driving adaptive threshold changes by decision result and process cost. First, we define the decision result cost and decision process cost for multi-class three-way decision-making. Second, we design a comprehensive importance measure of attributes under a specific class, which considers classification capability, decision result cost, and decision process cost, and use it to construct the granularity structure in the sequential multi-class three-way decision process. Then, we analyze the relationship between decision thresholds, misclassification rate, and deferred decision rate, and design an adaptive threshold selection method based on penalty function. Based on this, we develop a cost-sensitive sequential multi-class three-way decision model. Finally, experimental results show that, compared to existing models, the proposed model achieves lower decision cost while maintaining high accuracy.</p>