Abstract: To achieve rapid diagnosis of faults in cigarette draw resistance and ventilation rate detection equipment, a fault diagnosis method based on bivariate normal distribution and Mahalanobis distance was proposed, and applied to five artificially set fault types. First, the hypothesis that draw resistance and ventilation rate data conform to a bivariate normal distribution was validated using non-parametric tests and graphical analysis. Subsequently, a fault diagnosis model was established based on the confidence ellipses of the bivariate distribution of draw resistance and ventilation rate data, following the principle of Mahalanobis distance. Further, the contribution scores of outlier variables were analyzed to enable precise identification of fault sources. The results showed that: 1)The draw resistance and ventilation rate data in the training set conform to a bivariate normal distribution. 2)Under the 95% and 99% confidence levels, the proportions of outliers detected by the model in the training set are 4.81%–5.14% and 0.67%–0.93%, respectively, both within acceptable ranges. This indicates that the actual data distribution is more concentrated than the assumed bivariate normal distribution, showing good fitting. 3)The method shows high diagnostic sensitivity for severe main pipeline leakage (Type A) and slight main pipeline leakage (Type B), with maximum outlier detection rates of 82.08% and 53.59%, respectively. It also exhibits certain identification capability for upper sealing ring rupture (Type C) and middle sealing ring rupture (Type D), with maximum outlier detection rates of 12.40% and 16.53%, respectively; however, it fails to effectively identify lower sealing ring rupture (Type E). This research provided a theoretical foundation and technical support for the rapid diagnosis of equipment faults in cigarette production processes.