Megafauna is an informal term for large animals, especially very large animals like elephants and hippos. The threshold for an animal to be "megafauna" is variously defined as 44 kg (97 lb), 100 kg (220 lb), and 250 kg (551 lb). Sometimes, megafauna is divided into three categories: "small" (250–500 kg, 551-1102 lb), medium (500–1,000 kg, 1102-2204 lb), and large (over 1,000 kg/2204 lb). Intuitively, it seems that megafauna includes animals significantly larger than humans, including cows and horses.
The term megafauna is especially popular for describing the numerous large Pleistocene species that went extinct due to hunting or otherwise competing with humans — dire wolves, short-faced bear, mammoth, moa, saber-toothed tiger, etc. It may refer to Pleistocene (1.8 million to 10,000 years BP) animals that went extinct in the last couple million years or so — ground sloths, megalodon shark, "terror birds," etc., or more generally, any large animals in Earth's history.
The Pleistocene was an especially distinctive time for megafauna as it contained many animals that resemble, closely related to, or are in fact larger versions of surviving species today. For instance, although the mammoth was not especially larger than an elephant, it was closely related to it. The dire wolf was a larger version of today's gray wolf. The cave bear was a larger version of other living bears. There were giant eagles in New Zealand, 10 ft carnivorous birds in South America, even dog-sized rodents on Mediterranean islands and cow-sized hamsters in Uruguay!
Sometime around 2 million years ago, there was a massive dieoff in megafauna worldwide, followed by another pulse when humans spread across the world just 100,000 years ago. The latter extinction pulse is obviously caused by human hunting and competition, but the source for the former is unknown. Glaciation is sometimes pointed to as a culprit, but these species survived many cycles of glaciation before then without incident. Inter-species pandemics have been considered, but the extinction pattern is too long and drawn out to support this hypothesis. More research is needed to determine why much of the late Pleistocene megafauna disappeared.