It was discovered by George Alcock of Yaxley, Cambridgeshire, England at 4:35 UT on the morning of 25 March 1991.

He found it with 10×50 binoculars, and on that morning its apparent visual magnitude was 5 (making it visible to the naked eye).

[5] V838 Herculis declined from its peak brightness very quickly, fading by 2 magnitudes in less than three days, making it one of the fastest classical novae ever recorded.

The two stars are so close to each other that material is transferred from the donor to the white dwarf.

The eclipses were first detected a few weeks after the nova outburst, and they show the system's orbital period to be 7 hours, 8 minutes and 36 seconds as of 1991.