We have prehistorical findings that show traces of counting, e.g. the Ishango bone.
It is highly questionable, though, that counting alone can be considered "mathematics" (even though some scholars claim it to be).

If you (like the majority, including me) on the other hand consider some sort of "composition of elements (like addition or subtraction) as a method for solving a problem" as a precondition for classifying any source as mathematical, the oldest known texts stem from Babylonian times. Here again, the first signs of an understanding of a number concept are counting devices from the 5th millenium B.C., ususally some form of tokens. Later on we find mainly lists of things; those contain some sort of counting, again. The oldest known tablet, and hence the oldest known source in general, to contain content which is undoubtley [=undisputed] mathematical is W 19408,76 -- it contains two exercises for the calculation of areas. It was found in the area where the city of Uruk used to be and was produced sometime in the last quarter of the 4th millenium B.C.

For further reference you can confer to Eleanor Robson's Article Mesopotamian Mathematics in: The Mathematics of Egypt, Mesopotamia, China, India, and Islam: A Sourcebook. editet by V. Katz e.a.