The SAFARI (seismo acoustic fast‐field algorithm for range‐independent environments) wave propagation model is applied for the first time to low‐frequency atmospheric sound propagation and tested by means of synthetic and real‐world data. Problems experienced with the use of SAFARI are discussed in detail. Good agreement of the model output with theoretical predictions is achieved for two half‐spaces (air/ground). Available meteorological data up to 5 000‐m height are applied to the model in the next step in order to compute sound transmission loss versus range and receiver height. Results are given in transmission loss contour plots covering a field of 1 000 m in height and 30 000 m in range. The influence of typical sound velocity profiles including strong gradients close to the ground is investigated. It turns out that knowledge of meteorological data is most relevant for heights of up to about 200 m for sound propagation modeling within a field of interest as above. SAFARI is finally compared with a fast‐field program developed at the U.S. Army Construction Engineering Laboratory. This test yields very good agreement between both model predictions on a test case in which a real‐world sound profile is used. It has to be emphasized that all results are for very low frequencies where finite impedance ground effects are minimal. One would expect to see differences at higher frequencies. These encouraging results give rise to recommending the use of the models in a number of interesting applications such as various types of acoustic detection system and civil noise control.