The study of ancient DNA offers the possibility of following genetic change over time. However, the field is plagued by a problem which is unique in molecular biology--the difficulty of verifying results by reproduction. Some of the reasons for this are technical and derive from the low copy number and damaged state of ancient DNA molecules. Other reasons are the unique nature of many of the objects from which DNA is extracted. We describe methodological approaches with which these problems can be alleviated in order to ensure that results are scientific in the sense that they can be reproduced by others.