Volcanic lightning has intrigued observers through the centuries. Several likely processes have been proposed to explain the electrification of volcanic plumes, including quenching magma-water interactions, the fracturing or internal friction of fine grained ash, and the freezing of plume water at height. Scarce measurements of volcanic lightning have not been able to distinguish between proposed ideas. The Eyjafjallajökull volcanic eruption in Iceland in April–May 2010 may have revealed its charge mechanism. During its 39 days, the eruption went through a few phases while the conditions of the ambient atmosphere also changed, but at different times. The most surprising change in the lightning activity occurred on 11 May, with no obvious change in the physical eruption character or strength. During 3–10 May there was no lightning recorded by long-range networks, followed by intense activity 11–20 May. The change in lightning activity coincided with a change in the conditions of the ambient atmosphere. At this time the altitude of the isotherms for droplet freezing (about −20°C) dropped drastically below the plume top. Therefore, it appears that the atmospheric conditions around the plume were influencing or even controlling some of the lightning activity. The critical plume top temperature, which appears to have turned on and off the observed lightning activity during the Eyjafjallajökull eruption is estimated to be between −20° and −24°C. We conclude that a significant charge generation process of the observed volcanic lightning is probably analogous to processes in meteorological thunderclouds.