Glass transition temperature (T g ) is an important property for amorphous pharmaceutical formulations, because it is closely related to the storage stability. The T g of amorphous materials can usually be determined by calorimetry, but this technique cannot be applied to amorphous formulations containing polymers with widely distributed molecular weights. This is because these formulations often exhibit unclear changes in heat capacity at T g due to the glass transition occurring over a wide temperature range. Molecular dynamics (MD) simulations can be performed for an amorphous matrix model constructed using polymer molecules of a uniform molecular weight, in which the chemical structure of the repeated unit can be modified. If T g prediction is possible based on MD simulations, the dependence of T g on the polymer molecular weight as well as on the chemical structure of the repeated unit can therefore be elucidated, leading to the efficient development of polymer excipients with high T g values suitable for stable amorphous dosage forms. Furthermore, MD simulations can determine the dependence of fragility for polymer matrices on the polymer molecular weight and on the chemical structure of the repeated unit. Thus, it would be possible to estimate the fragility parameter of polymer matrices with widely distributed molecular weights, which usually cannot be determined from the heating-rate dependence of T g nor from the width of glass transition.MD simulations have been utilized to estimate the glass transition temperature (T g ) of amorphous synthetic polymers, 1,2) glass former saccharides and concentrated saccharide-water systems. [3][4][5][6][7][8] These studies suggest that MD simulations are useful in estimating the T g of amorphous materials. Our previous MD simulations with isomaltodecaose (a fragment of dextran) and a-glucose (the repeated unit of dextran) demonstrated that MD simulations can provide rational T g values that decrease upon hydration and increase with increasing fragment size, suggesting the usefulness of MD simulations.9) However, the T g obtained from MD simulations was not compared with experimentally determined T g , in consideration of the heating/cooling-rate dependence. Such comparison is necessary in order to evaluate the reliability of MD simulations. In this study, the T g of an isomalto-oligomer of relatively narrow molecular weight distribution was determined as a function of heating and cooling rates by differential scanning calorimetry (DSC). For the other angle of investigation, MD simulations were carried out with an amorphous matrix constructed from isomaltoheptaose, the molecular weight of which was close to the isomalto-oligomer investigated. The density of the isomaltoheptaose matrix was calculated as a function of temperature, and the T g of the matrix was estimated from a change in the slope of the density versus temperature profile. The T g estimates obtained at various cooling rates were compared with the experimental data in order to examine whether MD simulati...