The invention relates to a method for building a cigarette bitter taste index evaluation model based on a taste activity value. The method for building the model comprises the steps of 1) selecting a representative substance of which the alkaloids are the cigarette bitter taste and performing quantitative analysis by gas chromatography-mass spectrometry to obtain a free alkaloid detection result, 2) measuring the bitter taste threshold of each alkaloid in water medium by 3-AFC, calculating the alkaloid taste activity value (TAV) and identifying cigarette key bitter taste substances, 3) building the cigarette bitter taste index evaluation model based on the cigarette key bitter taste substances, and 4) performing fitting effect and model verification analysis on the cigarette bitter taste index and a human body sensory evaluation score by use of Origin 8.5 statistical software. A relatively ideal fitting effect of the built cigarette bitter taste index evaluation model and the human body sensory evaluation is achieved, objective evaluation on the cigarette bitter taste intensity is realized, and technical support can be provided for cigarette bitter taste mechanism revealing, bitter inhibitor development and cigarette bitter taste evaluation.