Our objectives were to compare the Couchman-Karasz-predicted glass transition temperature, T-CK, to the measured glass transition temperature of a mixture, T-gm, of model confectionary systems and develop an empirical correction to improve the accuracy of prediction. Results showed that the original Couchman-Karasz equation fit the data better than the modified equation; although both generally overestimated T-gm. Blends containing sorbitol had the largest Delta T-gm(CK), where Delta T-gm(CK) = T-CK - T-gm. While T-gm varied with composition, the increase in Tgm with decreasing moisture content was linear ((R-2) over bar = 0.984) and consistent across blends (4.5 +/- 0.9 degrees C/1% moisture, wb). The increase in T-gm with increasing cook temperature was best described by a polynomial model ((R-2) over bar = 0.998), but adequately described by a more generalizable linear model ((R-2) over bar = 0.979). Application of an empirical correction based on moisture content or cook temperature and T-CK of dry ingredients reduced the average Delta T-gm(CK) from 20.1 degrees C and 11.3 degrees C for modified and original equations, respectively, to <5.6 degrees C. (C) 2017 Elsevier Ltd. All rights reserved.