AbstractRecent guidelines released by the U.S. EPA define a worst‐case scenario as a release under stable atmospheric conditions defined as Pasquil‐Gifford stability class F. Unfortunately, very few tests at F stability have been available heretofore to provide a basis for models. Recent test data with propane releases by the German research organization TUV provide a set of 60 experiments conducted specifically to define the effects of atmospheric stability class on dispersion. Of these, 25 tests were at F stability. A comparable number were at each other stability class A through E. In addition 23 tests were at wind speeds under 1.5 m/s in stable atmospheres. This paper reports on adjustments made to our models based on these new data by reducing the originally‐postulated sensitivity to stability class. In spite of considerable scatter in the TUV data, particularly between two different types of propane analyzers, the model allows us to extract information by averaging over the