The method of calculating groundwater nitrate background level was explored in Liujiang Basin as a research object. Based on a comparison and analysis of methods adopted both in China and abroad, the authors firstly used an absolute concentration and milliequivalent percentile double factors method to eliminate outliers macroscopically, and then employed hierarchy clustering analysis combined with principal component analysis to investigate hydrochemical characteristics and identify anomaly category for further candidate selection. The last remaining data were examined for distribution type with the range of groundwater nitrate background level being determined by the cumulative frequency method. The results show that, although the absolute concentration and milliequivalent percentile double factors method cannot completely eliminate outliers, it helps to reduce abnormal information and subset for subsequent hierarchical clustering analysis identifying outliers. Hierarchical clustering analysis method takes advantage of identification of artificial and natural anomaly in that it emphasizes groundwater hydrochemical characteristics analysis. Compared with the mathematical methods, it is more efficient to identify outliers and more reasonable to calculate groundwater nitrate background level by using the two methods in combination. The outliers analysis shows that the high nitrate content in shallow groundwater is closely related to overuse of agricultural fertilizer and living sewage and faeces infiltration in Liujiang basin.