NONMEM. is a software widely used in the field of population pharmacokinetics and pharmacodynamics, mainly for related data analy-sis. In theory, it mainly establishes a parameterized model, combines the obtained data, and uses dif-ferent parameter estimation methods to estimate the parameters in the model, and then analyzes the data according to the model. This paper briefly introduces the representation of parameterized models in NONMEM., and from statistical theory, summarizes three commonly used parameter esti-mation methods under the condition that the ran-domization effect parameters η and .do not inter-act. For nonlinear mixed effects models, the rela-tionships among three parameter estimation meth-ods are given under special cases of addictive intra-individual models and proportional intra-individual models. In addition, through numerical experi-ments on data of four pharmaceutical companies on the change of ibuprofen drug concentration with time, the rationality of theory is further veri-fied in terms of calculation time and model predic-tion residuals.