Ill-conditioned problem which causes great harm to survey results is widely existed in survey measuring.As in the field of deformation monitor,insufficient measure data or additional parameters chosen would enlarge deviationbetween least squares estimation solution and true value; the results will be not stable.In this paper,on the basis of specific measure situation,the capabilities of regularization methods and direct methods for solving ill-conditioned problems are compared.According to experimental results,on the condition of serious ill-conditioned problems,the normal least squares estimation solution is inherently unreliable for the obtained results are away from true value.And the variation law is: result varies for constant error disturbance.If the error is large,the results disturbance is larger and vice versa.For regulation parameters,it varies for constant error disturbance,either.Besides,the results are not completely correspond with normal matrix condition number,for example,when the regulation parameter value is large,however the condition number is small,and the results are not as good as the tiny error disturbance results.So,reduce normal matrix condition number should be accompanied with lowering constant error disturbance.Finally,take 71505working face in Dongshan coal mine as an example,the application of regularization methods in deformation monitor data processing can be studied.