Abstract:Nonlinear mixed effects model (NLMEM) is the model in which both the fixed and random effects occur nonlinearly in the model function. First-order linearization algorithm (FO) and conditional first-order linearization algorithm (FOCE) are two commonly used linearization algorithms to calculate the parameters in NLMEM. We proposed an improved method for calculating random effects parameters based on FOCE algorithm in this study. We also analyzed and compared the three algorithms using height growth data set and simulation data sets. The results are: random effects parameters obtained from improved FOCE algorithm can more really reflected the individual random variations and also make a high efficient fit.