This study analyzes the basic cost factors in the automation of library catalogs, with a separate examination of the influence of typography on the cost of printed catalogs and the use of efficient automatic error detection procedures in processing bibliographic records. The utility of automated catalogs is also studied, based on data from a random sample of the shelf list of a medium-sized university library. An investigation of several large university and public libraries shows that the collections of mature libraries grow at a rate close to the rate of growth of the Gross National Product, and the significance of this relationship as a library management tool is discussed. Numerical methods for determining useful ratios of the size of library files and the number of entry fields in the structure of the record are given and applied to the use of machine-readable catalog data in the production of bibliographies. It is concluded that mechanization of the cataloging function is necessary and inevitable, and it is recommended that the machine-readable catalog data be used to produce printed book catalogs and special purpose bibliographies. Appended is a description of the selection of a random sample from the shelf list of the Fondren Library at Rice University. (Author/JB)