The classic economic account of the origins of money is that of Carl Menger. Menger described money as emerging out of markets where commodoties were bartered. Barter markets were inefficient, Menger explained, because a double coincidence of wants was required before a trade could be consumated. One way to solve this would be credit, but credit is a very imperfect institution for a variety of reasons. Another way to solve the problem is the intermediate commodity -- a commodity that everybody is willing to take or give in trade. This would start by being an otherwise useful commodity, but eventually it would be come to be valued more, or even entirely, for its monetary attributes (i.e. its attributes as an intermediate commodity, such as durability, divisability, the ability to be securely stored, etc.) Thus would money naturally emerge from the operations of a barter market.
Menger's theory is valuable in a number of ways. The coincidence of wants problem is a crucial idea as well as illustrating the Austrian economic ideas of time and subjective preferences. Menger's theory provides a good example of how economic institutions can emerge, it demonstrates that (contrary to modern legal definitions) money need not be created by any sort of government, and it suggests that the emergence of money long predated the invention of coinage. And Menger's theory is true insofar as a barter market would evolve in the way Menger describes. However, as an actual description of the origins of money it is almost surely wrong -- or alternatively, it is even more right than he could have known. This is because money, in the form of collectibles such as shells, predated low transaction cost commodity markets by tens of thousands of years.
Money (or, if you prefer to stick to modern legal definitions of money, proto-money which I call "collectibles") emerged far sooner than barter markets because the double coincidence of wants problem occurs not only in barter exchanges, but in other transactions that were as or more important than barter to hunter-gatherer societies: paying tribute, paying legal fines, bride price, and mortuary distribution (inheritance). Furthermore, barter exchanges in these societies usually resembled far more a bilateral monopoly than an efficient market exchange. Use of shell jewelry or similar objects in hunter-gatherer and neolithic societies for these purposes is nearly universal. Nevertheless, in all these kinds of transactions a double (or more) coincidence of wants and events (such as death, marriage, or legal judgment) is required if they are done in-kind rather than with objects with good monetary properties; the best such objects available were usually shells, often strung on necklaces providing secure storage and divisibility.
Furthermore, these appeared long enough ago (mature shell jewelry is found now back to 75,000 B.P.) that the (begging) explanation "people collected shells and made jewelry for pleasure" may be backwards -- humans may well have evolved the pleasure of collection because of the evolutionary benefits of greater cooperation. Lowered transaction costs for all these kinds of transactions meant greater familial, political, legal, and economic cooperation -- i.e. the enhancement of kin altruism as well as reciprocal altruism and the mitigation of aggression.
This is spelled out in detail in my Shelling Out: The Origins of Money
