Abstract: How to make a reasonable criterion to accept or reject measurements is a valuable research topic.The current fixed-tolerance-method,which is to reject the outliers and accept all the others,in fact,has quite a few disadvantages.Here,this paper tries to enrich this method with a control graph for discriminations of abnormal levelling measurements within tolerance,three criterions are summarized as follows: Criterion 1,although the points are within tolerance,the case in which a certain number successive points are close to the critical lines is to be judged abnormal; Criterion 2,although the points are within tolerance,the case in which over 5 successive points distribute at one side of the central line is to be judged abnormal; Criterion 3,although the points are within tolerance,the case in which over 5 suc-cessive points centre around the vicinity of the central line is to be judged abnormal. The above criterions are based on the principle of statistical test.This control-graph-method is so straightforward that we can decide in time what measurements are normal and what are not.For field levelling,this method is not only economical and feasible, but also of benefit to improving the quality of levelling. However,there are other problems to be solved.Such as,how to determine whe-ther the points are close to the critical lines or the central line,what the factors re-sulting in the abnormality exactly are,etc.these need further studying.Only so,is the method for the quality control of surveying being perfected.