Data storage and retrieval expenses represent a significant portion of the cost of operating a management information system. The study focuses on the question of how much data already collected need be stored for future use, and on methods of reducing the quantity of data without necessarily reducing the information content. Several linear interpolation and least squares methods are explored for achieving data reduction, using as a means of illustration twenty-three different types of hydrologic records. Discussed also is the value of the data, the desired accuracy needed for various water resources studies, and the costs of data reduction as compared to data storage and retrieval.