Glucose minimal model parameters are commonly estimated by applying weighted nonlinear least squares to each individual subject's data. Sometimes, parameter precision is not satisfactory, especially in "data poor" conditions. In this work, the use of population analysis through nonlinear-mixed effects models is evaluated and its performance tested against the parameter estimates obtained by the standard individual approach through weighted nonlinear least squares. In particular, we compared the performance of two likelihood approximation methods to estimate nonlinear mixed-effects model parameters, i.e. the first-order conditional estimation (FOCE) and the Laplace approximation (Laplace) methods. The results show that nonlinear mixed-effects population modeling using the FOCE approximation can be successfully used in order to accurately estimate individual minimal model parameters.