NONMEM is one of the most used software in the pharmaceutical field to handle population pharmacokinetic (PK) - pharmacodynamic (PD) analysis. Numerous estimation algorithms are implemented in the latest version (NONMEM 7.4.4), making the choice of algorithm for NONMEM users difficult. This study aims to investigate the performance of six estimation algorithms, two classical methods (FOCEI and LAPLACE) and four expectation maximization (EM) methods (IMP, IMPMAP, ITS, and SAEM) with respect to parameter estimation and runtime. This study involved the re-estimation of 26 previously published models from the Pharmacometrics group of Uppsala University (17 continuous and 9 categorical) in addition to one continuous model that was obtained from the ddmore foundation repository website. MU- referencing has been investigated as much as possible for the used models. In this study, the estimation algorithms that showed the best performance from the perspective of the OFV were the FOCEI for the continuous models and the LAPLACE for the categorical models, as both algorithms resulted in a higher number of successfully converged models. In addition, both algorithms required less time in most used models compared to the other algorithms.