The bioavailability and bioequivalence between drug products has become an important subject in drug development. The average bioavailability of the test (T) and the reference (R) products is currently specified in the FDA guidelines to be used for assessing the bioequivalence of the drug products. However, it has been recognized that the safety for the substitution of a reference drug product with a test drug product in patients, whose concentration may have been titrated to a steady efficacious and safe level, could be a concern. Therefore, it is suggested that individual bioequivalence within each subject be assessed to assure the safety of the drug switchability. This paper examines the statistical properties of TIER procedure that Anderson and Hauck (1) proposed to assess individual bioequivalence. It is shown that Anderson and Hauck's procedure could be improved by imposing some distribution assumption such as lognormal distribution for assessment of individual bioequivalence. This paper also compares the relative performance of the individual bioequivalence based on TIER procedure and the average bioequivalence based on two one-sided tests procedure suggested by Schuirmann (2). The relationship between equivalence limits for the improved TIER procedure and average bioequivalence is also examined.