Highly variable drugs (within-subject variability greater than 30%) have been difficult to meet current regulatory acceptance criteria using a reasonable number of study subjects. In this study, we reviewed previous studies presenting alternative approaches for bioequivalence evaluation of highly variable drugs, and focused on an approach for widening the bioequivalence acceptance limits using within-subject variability. We discussed the suggested five solutions for highly variable drug including the deletion of $C_{max}$ of the bioequivalence criteria, direct expansion of bioequivalence limit, multiple dose studies in steady state, bioequivalence assessment on the metabolite, add-on study, and widening the bioequivalence acceptance limits based on reference variability. The methods for widening of bioequivalence limits based on reference variability are scaled average bioequivalence containing within-subject variability on reference drug (${sigma}_{WR}$), population bioequivalence derived from total variability on reference drug (${sigma}_{TR}$) and test drug (${sigma}_{TT}$), and individual bioequivalence derived from subject by formulation interaction variability (${sigma}_D$) and within subject variability on reference drug (${sigma}_{WR}$) and test drug (${sigma}_{TR}$). To apply these methods, the switching variability (${sigma}_0$) will have to be set by the regulatory authorities. The proposals of bioequivalence evaluation approach for the highly variable in Korea are presented for both of new drug and reevaluation drug.