Using ellipsometry, we characterized the nanoconfinement effect on the glass transition temperature (T (g)of supported polystyrene (PS) films employing two methods: the intersection of fits to the temperature (Tdependences of rubbery- and glassy-state thicknesses, and the transition mid-point between rubbery- and glassy-state expansivities. The results demonstrate a strong effect of thickness: T(g) (bulk) - T(g)(23 nm) = 10 degrees C. The T -range needed for accurate measurement increases significantly with decreasing thickness, an effect that arises from the broadening of the transition with confinement and a region below T (g) where expansivity slowly decreases with decreasing T . As determined from expansivities, the T (g) breadth triples in going from bulk films to a 21-nm-thick film; this broadening of the transition may be a more dramatic effect of confinement than the T (g) reduction itself. In contrast, there is little effect of confinement on the rubbery- and glassy-state expansivities. Compared with ellipsometry, T (g) 's from fluorescence agree well in bulk films but yield lower values in nanoconfined films: T (g)(bulk) - T (g)(23 nm) = 15( degrees ) C via fluorescence. This small difference in the T (g) confinement effect reflects differences in how fluorescence and ellipsometry report "average T (g) " with confinement. With decreasing nanoscale thickness, fluorescence may slightly overweight the contribution of the free-surface layer while ellipsometry may evenly weight or underweight its contribution.