The results of a study of eight different key-to-address transformation methods applied to a set of existing files are presented. As each method is applied to a particular file, load factor and bucket size are varied over a wide range. In addition, appropriate variables pertinent only to a specific method take on different values. The performance of each method is summarized in terms of the number of accesses required to get to a record and the number of overflow records created by a transformation. Pccularities of each method are discussed. Practical guidelines obtained from the results are stated. Finally, a proposal for further quantitative fundamental study is outlined. © 1971, ACM. All rights reserved.