A black swan event is one that is highly improbable (and unforeseen) that nonetheless occurs and has a significant impact.
The term was coined by Nicholas Nassim Taleb, whose book of the same name, argues that markets, history and life in general are dominated by black swan events. Black swans include wars, technological breakthroughs, and anything else that is not anticipated.
The concept of black swans is closely related to that of fat tails, but the crucial difference is that there is no reason per se that fat tails cannot be quantified and incorporated into models.
Black swans cannot be anticipated or quantified, therefore their risk cannot be Incorporated into models. A black swan is by definition both significant and unprecedented. There is no way of measuring the probability of the unprecedented, and black swans are too significant to be simply ignored.
Taleb is particularly contemptuous of the use of the normal distribution to measure risk, although his complete dismissal of it in all applications is too sweeping: it is very clearly useful in many areas.
The term black swan refers to the belief in Europe that all swans were white, until the unexpected discovery of black swans in Australia.