The objective of the present study was to develop an empirical cold hardiness model applicable to several taxa of deciduous trees. Cold hardiness expressed as lowest survival temperature of Acer rubrum, Betula nigra, Liquidambar styracifiua, Fraxinus pennsylvanica, Prunus serotina and Quercus alba was evaluated at approximately weekly intervals during the winters of three consecutive years. Plant samples and meteorological data were collected from Georgia Experiment Station, Griffin, Georgia. Maximum, minimum and average temperatures, hourly chill and heat accumulation. day length and time of year were used as input variables for model development. The statistical method of stepwise procedure of regression analysis was employed to select variables for the model. Based on the assumption that model components should be the same for all taxa included in this study and all three winters, the following independent model variables were selected as valid inputs: day length, number of accumulated hours with temperature above 20°C and number of accumulated hours with temperature below 10°C. Equation coefficients of species-specific models were determined for each species. Cold hardiness predictions were compared to actual observations for each species. The model components were interpreted as representing two processes: (1) internally regulated and independent of actual temperature, and (2) externally regulated and dependent on the amount of accumulated chill or heat. The model allowed for comparisons of cold hardening and dehardening between the studied taxa and between years.