The objective of this study was to characterize and predict the air-temperature profiles in forest canopy using empirical models and Landsat-7 Enhanced Thematic Mapper Plus (ETM+) thermal infrared (TIR) data. For each hour of the day, empirical models were developed using the air temperatures measured in a mixed forest in Tolland, Connecticut during the summer of 2001. Air-temperature profiles measured on eight randomly selected days over the study period were used to evaluate the models' performance. The results show high agreement between model-predicted temperatures and field observations (R2=0.96 and RMSE=0.5°C). Forest canopy-surface temperature was estimated from Landsat ETM+ TIR data and then applied on the empirical model at 10:30 to predict air temperatures below the forest canopy. Compared with field-measured air temperatures on the same day, good agreement (R2=0.80 and RMSE=0.7°C) was obtained between predicted and observed air-temperature profile in this mixed forest. This study implies that Landsat TIR data combined with empirical models have the potential to characterize and further predict the air-temperature profile in forest canopies.