This paper describes the method of radioactivity measurement to eliminate the outliers in experiment. The value of radon concentration can cause deviation in the process of transmission because the FH-463 is interfered in the process of measurement by external factors such as electromagnetic waves etc. It makes measurements vary greater than the Expectations. And it can only process 10% or less Outlier when using Grubbs’ criterion directly. In order to meet the Radon Concentration for inspection, the paper makes some necessary improvement on the criterion to adapt itself and estimates the sample data with linear regression. The experimental results show that the method is very convenient to reject the data with deviation more effective. The method in this paper can also be used in other fields.