In order to determine the relationship of the socioeconomic variable sets of education, general characteristics, and income and housing to geographic variations in the collection size of academic and public libraries in the United States, data for the 50 largest Standard Metropolitan Statistical Areas (SMSAs) are analyzed using frequency distribution, correlation analysis, and regression analysis. Two hypotheses are tested: (1) the socioeconomic character of the SMSAs will relate to the collection size of both academic and public libraries, and (2) academic library collection size reveals greater correspondence with these socioeconomic characteristics than public library collection size. Analysis of data lends support to the first hypothesis but refutes the second, suggesting that the public library model is more strongly influenced by population. Supporting material is presented in 18 tables and 10 figures, and a bibliography is provided. An alphabetical listing and a listing ranked by population for the 50 largest SMSAs in 1973, frequency distributions, and regression tables are appended. (KM)