Flash‐level comparisons between the Geostationary Lightning Mapper (GLM; primarily from GOES‐16), the U.S. National Lightning Detection Network (NLDN), and the Global Lightning Dataset GLD360 have been done at quasi‐climatological scale and in smaller samples of both severe and nonsevere thunderstorms. A small sample of data from GOES‐17 has also been analyzed. These comparisons show that the total lightning detection efficiencies of the GLM instruments drop off within about 2,000 km of the edges of their fields of view, particularly over terrestrial areas. In severe storms, low detection efficiency by GLM is clearly associated with very high midaltitude reflectivity, consistent with the idea that large multiple scattering path lengths, together with absorption of the near‐infrared signals by water (in all of its phases), depresses the GLM detection efficiency. This effect appears to be coupled with additional factors, including time of day, flash energetics, and the incident angle at the GLM sensor. The “lightning jump,” the sought‐after signature of severe storms in lightning observations, tends to be poorly correlated between NLDN and GLM unless the storms are fairly isolated, not close to the edge of the GLM field of view, and have moderate midaltitude reflectivity.