Abstract: In order to decrease test cost and increase test efficiency based on ISO 12863 standard method, a sample number-adjustable assessment method for ignition propensity evaluation was developed by combining a sampling scheme and a statistics algorithm. In addition, theoretical reasons for large deviations of the test result index were analyzed via binomial distribution analysis on the full-length burn percentage (P-FLB). The results showed that: 1) The accuracy and repeatability of the improved method were consistent with the ISO standard method. 2)P-FLB was the highest point in the confidence interval of the test results, however, its confidence was still low, thereby leading to a theoretical maximum deviation of around 10%-20%. 3) When the number of full-length burn cigarettes was lower (between 0-4 cigarettes), the ISO method was more stable; when the number was higher(between 5-10 cigarettes), this new method was more stable. When the number was more than 6, there were theoretically potential risks in fire safety performance of the cigarettes based on this test principle. 4)The test results of 80 cigarette samples with lower ignition propensity indicated that both the average test time and the Whatman filter paper consumption of this method were reduced by 59.65% comparing with the original ISO method.