In this paper we report that scaled nucleation theory (SNT) can describe moderately well the observed nucleation behavior of a significant number of refractory materials if a more appropriate value of a quantity commonly referred to as the excess surface entropy is used. With the availability of more reliable critical point and liquid property data, we are better able to calculate this quantity and we find that for refractory materials it can be as small as one half to one third the quantity traditionally used in its approximation. As a result of using more accurate values, we find considerably better agreement between SNT and experiment than what was originally determined. We also explain why using surface tension slope information to determine the excess surface entropy can lead to substantial errors in the SNT supersaturation prediction.