An in vivo bioequivalence (BE) study is an important component in the submission of abbreviated new drug applications (ANDAs) to support the conclusion of therapeutic equivalence between a generic drug product and reference listed drug. There are standard study designs and statistical analysis method for bioequivalence evaluation. However, there are challenges in performing standard BE studies and analyses in some circumstances. One challenge is to determine the BE of highly variable drugs that are defined as drugs with within-subject variability exceeding 30%. Another challenge is BE evaluation for drugs with a long half-life. Population pharmacokinetic analysis based on nonlinear mixed-effects models (NLMEM) has been widely used in new drug development and evaluation. It is flexible to handle a variety of study designs and often have higher power. As a result, model-integrated analysis serves as a promising strategy to solve challenges in BE evaluation for certain drugs. In this project, we plan to develop model-integrated methods to evaluate bioequivalence for certain drugs, e.g. those with high variation and/or long half-life. Simulation study will be performed to evaluate the developed method in the aspect of type I error and power. Strategy of model building and validation for model-integrated BE methods will be explored and illustrated using real data. In addition, the R package bemod that we developed previously will be updated to include resultant method expansions and improvements.