But this is not only a medieval or ancient problem for historians.
I finished watching Black Hawk Down this morning and decided to read a bit about the First Battle of Mogadishu. I was very surprised by what I found about the casualty roll, even though this event happened in 1993 (hardly ancient history!). Quoted from Wiki:
Somali casualty figures are unknown, but American estimates are that between 1,000 and 1,500 Somali militiamen and civilians lost their lives in the battle, with injuries to another 3,000-4,000. However, the International Committee of the Red Cross estimated 200 Somali civilians killed and several hundred wounded in the fighting. The book Black Hawk Down: A Story of Modern War estimates more than 700 Somali militiamen dead and more than 1,000 wounded, but the Somali National Alliance in a Frontline documentary on American television acknowledged only 133 killed in the whole battle. The Somali casualties were reported in The Washington Post as 312 killed and 814 wounded.
Here we have various sources, on both 'sides' giving wildly different casualty figures. As different as a factor of 10, which I find particularly interesting because the difference in French death count at Agincourt also varies by a factor of 10 in the various medieval sources. It seems this is not only a medieval problem.