Objective When the competition risk exists,the method based on restricted mean time lost(RMTL)has fewer model assumptions and more intuitive explanation.The intergroup effect size is RMTL difference(RMTLd).The corresponding hypothesis test is based on large samples,but the performance effect under small samples is unknown.Methods In this study,we explore the performance of RMTLd under small samples,and develop several variable conversion methods of RMTL to improve the statistical performance,and evaluate their type I error and power in different situations by Monte-Carlo simulation.Results In the case of small samples,the original method of RMTLd test has the phenomenon of type I error expansion,while the logic transformation method,which is one of the four transformation methods,can maintain good statistical performance.Conclusion When analyzing small sample competitive risk data,it is recommended to use the logical transformation of RMTL for statistical analysis.