A radiation balance method and a modified radiation balance method were used to determine hourly and daily values of atmospheric thermal radiation on three clear and three cloudy days during February and March 1970 at Phoenix, Arizona. The radiation balance was used over surfaces of bare soil, Bermuda grass, and open water; whereas the modified radiation balance was used over the open water surface only. The modified radiation balance differed from the pure radiation balance method in that net radiation was calculated as the residual in the energy equation rather than measured.The values of atmospheric thermal radiation thus obtained were compared to valuesobtained from measurements of incoming, all‐wave radiation minus incoming solar radiation.From this analysis it appeared that the best alternative to the semi‐direct measurementof atmospheric thermal radiation was the radiation balance over the bare soil surface.Results from this method were of sufficient accuracy, even on an hourly basis, for mostresearch studies, and were even better on a daily basis. Radiation balances over grass andwater surfaces were somewhat less satisfactory, but still better than the modified radiationbalance met