We performed an experiment to test the suggestion that the ultrasound radiated by certain severe storms is caused by lightning. During the 1972 storm season we recorded at Boulder, Colorado, the rate and arrival direction of both VLF atmospherics and infrasound from severe thunderstorms in the midwestern United States. If infrasound were caused by lightning, we should have observed a good agreement in direction and time between the radio and acoustic emissions of lightning, within observational uncertainty. Fewer than half of the infrasound events showed such agreement with electromagnetic emissions. Those agreements can be attributed to noncausal coincidence. We argue that the correlation should be much higher if the infrasound emissions were caused by lightning. Some detailed case studies illustrate the differing phenomenologies of the emissions: for example, they show that the infrasound is probably emitted during an earlier stage of storm growth than that usually associated with lightning.